JP4631807B2 - Air-fuel ratio control device for internal combustion engine - Google Patents

Description

  The present invention relates to an air-fuel ratio control apparatus for an internal combustion engine, and more specifically, a canister that adsorbs evaporated fuel generated in a fuel tank, and purge gas purged from the canister together with fuel injected from a fuel injection valve into an intake passage. The present invention relates to an air-fuel ratio control device for an internal combustion engine that can be used as fuel for the internal combustion engine.

  An internal combustion engine for a vehicle is provided with a canister that adsorbs and stores evaporated fuel generated in a fuel tank. The evaporated fuel stored in the canister is purged from the canister using the negative pressure in the intake passage during operation of the internal combustion engine. The evaporated fuel purged from the canister is diluted with air introduced from the air hole of the canister to become purge gas, supplied to the intake passage, and burned in the combustion chamber. In such an internal combustion engine, in order to always maintain good control accuracy of the air-fuel ratio, the fuel injection amount is reduced according to the supply amount of the purge gas during the purge of the evaporated fuel from the canister.

  The influence of the supply of the purge gas on the air-fuel ratio varies depending on the fuel concentration of the purge gas. For this reason, it is important to accurately grasp the fuel concentration of the purge gas in order to maintain the air-fuel ratio accurately in the vicinity of the target air-fuel ratio. Conventionally, in air-fuel ratio control of an internal combustion engine, a vapor concentration correction coefficient corresponding to the fuel concentration of purge gas is learned from a steady deviation of the feedback correction coefficient from its reference value during execution of purge control. The feedback correction coefficient is a correction coefficient to be applied to the fuel injection amount in order to achieve the target air-fuel ratio, and is negative when the air-fuel ratio (exhaust air-fuel ratio) corresponding to the exhaust gas sensor signal is richer than the target air-fuel ratio. When the exhaust air-fuel ratio is leaner than the target air-fuel ratio, it is set to a positive value.

The vapor concentration correction coefficient is used to set the purge correction coefficient. The purge correction coefficient is a correction coefficient to be applied to the fuel injection amount in order to cancel the influence of the purge gas on the air-fuel ratio, and the vapor concentration correction coefficient is multiplied by the purge rate, that is, the ratio of the purge gas flow rate to the intake air flow rate. It is obtained with. As learning of the vapor concentration correction coefficient progresses, the steady deviation between the feedback correction coefficient and the reference value accompanying the supply of the purge gas is replaced by the purge correction coefficient, and the feedback correction coefficient increases or decreases around the reference value. To be able to.
Japanese Patent Laid-Open No. 2003-332052 Japanese Patent Laid-Open No. 5-113140 JP 2005-42659 A

  By the way, in the internal combustion engine, a part of the fuel injected from the fuel injection valve adheres to the wall surface of the intake passage and is not used for the combustion immediately thereafter. During steady operation of the internal combustion engine, the amount of fuel attached to the wall surface is substantially constant because the new amount of adhesion and the amount of desorption are balanced, and as a result, the amount of fuel used for combustion coincides with the fuel injection amount. On the other hand, during the transient operation of the internal combustion engine, the amount of fuel adhering to the wall surface changes as the intake air amount and the fuel injection amount increase or decrease. When the amount of fuel adhering to the wall surface increases, the amount of fuel used for combustion becomes smaller than the fuel injection amount, leading to a lean air-fuel ratio. On the other hand, in the process in which the amount of fuel adhering to the wall surface decreases, more fuel than the fuel injection amount is used for combustion, leading to rich air-fuel ratio. For this reason, during the transient operation of the internal combustion engine, the corrected fuel amount is calculated based on the engine rotational speed, the load, and the coolant temperature so that the change in the fuel amount adhering to the wall surface is offset, and the fuel injection amount increase correction or Weight loss correction is applied.

  Some of the purge of evaporated fuel from the canister greatly affects the adhesion of fuel to the wall surface. Since evaporative fuel is less likely to adhere to the wall surface than liquid fuel injected from the fuel injection valve, the higher the ratio of evaporative fuel in the total amount of fuel supplied into the combustion chamber, the same operating conditions (engine speed) , Load, cooling water temperature), the amount of fuel adhering to the wall surface decreases. Further, the amount of purge of evaporated fuel not only affects the amount of fuel adhering to the wall surface and the amount of fuel desorbed from the wall surface, but also the stable amount of fuel adhering to the wall surface of the intake passage, that is, It also affects the amount of fuel attached to the stable wall. Specifically, when the purge amount of the evaporated fuel is large, the amount of fuel adhering to the stable wall surface is smaller than when the purge amount is small because the fuel injection amount from the fuel injection valve is small.

  For this reason, when the purge amount is reduced, the fuel adheres to the wall surface as the amount of fuel adhering to the stable wall increases, and until the amount of fuel adhering to the wall surface reaches the amount of fuel adhering to the stable wall surface, The air-fuel ratio becomes lean due to the shortage of the supplied fuel. On the other hand, when the purge amount is increased, the desorption of fuel from the wall surface proceeds as the amount of fuel adhering to the stable wall surface decreases, and until the amount of fuel adhering to the wall surface reaches the amount of fuel adhering to the wall surface, The amount of fuel supplied becomes excessive and the air-fuel ratio becomes rich.

  In an internal combustion engine, air-fuel ratio feedback control is performed based on a signal from an exhaust gas sensor. Therefore, even if a deviation occurs in the air-fuel ratio due to a change in the amount of fuel adhering to the wall surface, the target air-fuel ratio will eventually be corrected by correcting the fuel injection amount. Converge to. However, if the vapor concentration correction coefficient is learned in the process as described above, even the deviation of the feedback correction coefficient due to the change in the amount of fuel attached to the wall surface is reflected in the vapor concentration correction coefficient. The vapor concentration correction coefficient, which is the learned value, will affect not only the present time but also in the future.If the change in the amount of fuel adhering to the wall is mislearned as the change in the fuel concentration of the purge gas, The accuracy of air-fuel ratio control will be lowered for a while after that.

  The problem related to mislearning of the fuel concentration due to the change in the amount of fuel attached to the wall surface becomes particularly noticeable at the time of cold start where the fuel wall surface adhesion increases. Further, this problem is always a problem that can occur not only when the purge amount increases / decreases, but also when the adhesion state of the fuel adhering to the wall surface becomes unstable, such as when the fuel is cut or returned from the fuel increase amount.

  The present invention has been made to solve the above-described problems, and is an internal combustion engine that can prevent erroneous learning of the fuel concentration of the purge gas in a situation where the adhesion state of the fuel adhered to the wall surface becomes unstable. An object is to provide an air-fuel ratio control device.

In order to achieve the above object, a first invention is an air-fuel ratio control apparatus for an internal combustion engine,
A fuel injection valve for injecting fuel into the intake passage of the internal combustion engine;
A canister that adsorbs the evaporated fuel generated in the fuel tank;
A purge mechanism for flowing purge gas containing evaporated fuel from the canister into the intake passage;
An exhaust gas sensor disposed in the exhaust passage of the internal combustion engine and outputting a signal corresponding to the air-fuel ratio of the exhaust gas;
A feedback correction coefficient for calculating a correction coefficient (hereinafter referred to as feedback correction coefficient) to be applied to the fuel injection amount in order to realize the target air-fuel ratio in accordance with the difference between the target air-fuel ratio and the air-fuel ratio corresponding to the exhaust gas sensor signal. A calculation means;
A concentration learning means for learning the fuel concentration of the purge gas based on a steady deviation from a reference value corresponding to the target air-fuel ratio of the feedback correction coefficient during operation of the purge mechanism;
Purge correction coefficient calculating means for calculating a correction coefficient (hereinafter referred to as a purge correction coefficient) to be applied to the fuel injection amount in order to cancel the influence of the purge gas on the air-fuel ratio based on the learned value of the fuel concentration;
Fuel injection amount calculating means for calculating the fuel injection amount of the fuel injection valve by reflecting the feedback correction coefficient and the purge correction coefficient in the basic fuel amount;
Learning method correction means for acquiring a parameter related to the stability of the adhesion state of the fuel adhering to the wall surface adhering to the intake passage, and correcting the fuel concentration learning method by the concentration learning means according to the value of the parameter; ,
It is characterized by having.

According to a second invention, in the first invention,
The concentration learning means is configured to update the learning value when a deviation of the feedback correction coefficient from the reference value exceeds a predetermined dead band,
The learning method correcting means expands the dead zone according to the low stability of the fuel adhesion state determined from the parameter value.

According to a third invention, in the second invention,
The learning method correcting means is characterized in that either the upper limit value or the lower limit value of the dead band is changed according to the direction of change in the amount of fuel attached to the wall surface determined from the value of the parameter.

According to a fourth invention, in the first invention,
The learning method correcting means reduces the learning value update amount in accordance with the low stability of the fuel adhesion state determined from the parameter value.

According to a fifth invention, in the first invention,
The concentration learning means is configured to learn a fuel concentration based on a smooth value obtained by smoothing a deviation of the feedback correction coefficient with respect to the reference value in a time direction,
The learning method correcting means increases the smoothing coefficient used for smoothing according to the low stability of the fuel adhesion state determined from the parameter value.

According to a sixth invention, in the first invention,
The learning method correcting means extends the update period of the learning value according to the low stability of the fuel adhesion state determined from the parameter value.

According to a seventh invention, in the first invention,
The learning method correcting means suspends updating of the learning value when it is determined from the parameter value that the stability of the fuel adhesion state is low.

In an eighth aspect based on the seventh aspect,
The learning method correcting means restarts the updating of the learning value after a lapse of a predetermined period when it is determined from the parameter value that the fuel adhesion state is stable.

According to a ninth invention, in the first invention,
The learning method correcting means calculates a correction value for canceling the influence of a change in the amount of fuel adhering to the wall surface on the feedback correction coefficient based on the value of the parameter, and the feedback correction coefficient used for learning the fuel concentration The correction value is reflected in.

A tenth invention is the ninth invention,
The concentration learning means is configured to update the learning value when a deviation of the feedback correction coefficient from the reference value exceeds a predetermined dead band,
When the deviation is positive, the learning method correction means reflects the added value set to be equal to or higher than the upper limit value of the dead band in the feedback correction coefficient used for learning the fuel concentration, and when the deviation is negative Is characterized in that a subtraction value set below the lower limit value of the dead band is reflected in the feedback correction coefficient used for learning the fuel concentration.

In an eleventh aspect based on the tenth aspect,
The learning method correcting means is characterized in that, in a situation where it is determined that the amount of evaporated fuel in the fuel tank is small, the addition value or the subtraction value is reflected in the feedback correction coefficient.

In a twelfth aspect based on the tenth or eleventh aspect,
The learning method correcting means is characterized in that the added value is reflected in the feedback correction coefficient after the accumulated purge amount from the start of purge or a parameter related thereto reaches a predetermined value.

A thirteenth aspect of the present invention is any one of the first to twelfth aspects of the invention,
The learning method correcting means calculates a ratio of a fuel injection amount to a total fuel amount used for combustion, and acquires a deviation between the ratio and a smooth value obtained by smoothing the ratio in the time direction as the parameter. It is characterized by that.

In a fourteenth aspect of the invention, any one of the first through twelfth aspects of the invention
The learning method correcting means acquires the cooling water temperature of the internal combustion engine as the parameter.

In a fifteenth aspect of the invention, any one of the first to twelfth aspects of the invention,
The learning method correcting means acquires the elapsed time after returning from the fuel cut as the parameter.

In a sixteenth aspect of the present invention based on any one of the first to twelfth aspects of the invention,
The learning method correcting means acquires, as the parameter, a deviation between an increase coefficient reflected in a basic fuel amount for increasing the fuel injection amount and a smooth value obtained by smoothing the increase coefficient in the time direction. It is a feature.

  When the air-fuel ratio changes with the attachment of fuel to the wall surface or the desorption of fuel from the wall surface, the change in the air-fuel ratio is reflected in the feedback correction coefficient. Since the fuel concentration of the purge gas is learned based on a steady deviation from the reference value of the feedback correction coefficient, the change in the amount of fuel adhering to the wall is erroneously regarded as the change in the fuel concentration of the purge gas with the normal learning method. Learned. In this regard, according to the first aspect of the present invention, the fuel concentration learning method is modified according to the stability of the adhesion state of the wall surface adhering fuel adhering to the intake passage. Even when the fuel adhesion state is unstable, it is possible to prevent erroneous learning of the fuel concentration due to the fuel adhesion state.

  According to the second invention, the dead zone in which the learning value is not updated is expanded according to the low stability of the fuel adhesion state, thereby suppressing the influence of the change in the amount of fuel adhered to the learning value on the learning value. be able to.

  In particular, according to the third aspect, since the expansion direction of the dead zone is limited to the direction of change in the amount of fuel adhering to the wall surface, the learning value is updated as usual in the direction not affected by the change in the amount of fuel adhering to the wall surface. And a decrease in learning accuracy can be suppressed.

  According to the fourth aspect, the amount of update of the learning value is reduced according to the low stability of the fuel adhering state, so that the influence of the change in the wall surface adhering fuel amount on the learning value can be suppressed.

  According to the fifth aspect, the smoothing coefficient is increased in accordance with the low stability of the fuel adhesion state, so that the change in the amount of fuel adhered to the wall surface is reflected in the smooth value used for learning the fuel concentration. The speed can be reduced, and as a result, the influence of the change in the amount of fuel attached to the wall surface on the learning value can be suppressed.

  According to the sixth aspect of the present invention, the learning value update period is extended according to the low stability of the fuel adhering state, so that the influence of the change in the wall surface adhering fuel amount on the learning value can be suppressed.

  According to the seventh aspect, the update of the learning value is interrupted when the stability of the fuel adhering state is low, so that the influence of the change in the wall surface adhering fuel amount on the learning value can be suppressed.

  In particular, according to the eighth aspect of the invention, the learning value update is resumed after a predetermined time has elapsed after the fuel adhesion state is stabilized in consideration of the delay in transporting the gas from the intake passage to the exhaust gas sensor. The influence of the change in the amount of fuel adhering to the wall surface on the learning value can be more reliably suppressed.

  According to the ninth aspect, since the fuel concentration learning is performed based on the feedback correction coefficient in which the influence of the change in the amount of fuel attached to the wall surface is eliminated, even if the fuel adhesion state is unstable. The fuel concentration can be learned with high accuracy.

  In particular, according to the tenth invention, the feedback correction coefficient is increased in the direction corresponding to the deviation of the air-fuel ratio by reflecting the addition value or the subtraction value. By learning the fuel concentration using the padded feedback correction coefficient, the learning speed can be increased and the learning accuracy of the fuel concentration can be increased early.

  Further, according to the eleventh aspect of the invention, in the situation where the amount of evaporated fuel is large and the fuel concentration of the purge gas is unstable, the feedback correction coefficient is not increased and the learning update amount is calculated according to the normal deviation. It is possible to suppress the influence of the fuel evaporation status in the tank on the learning accuracy of the fuel concentration.

  According to the twelfth aspect, the feedback correction coefficient is increased in the lean direction by reflecting the added value. The fuel concentration is learned using the feedback correction coefficient padded in the lean direction, so that the purge of evaporated fuel from the canister proceeds and the fuel concentration of the purge gas is expected to decrease. The learning speed of the fuel can be increased, and the fuel concentration learning accuracy can be increased early.

  According to the thirteenth aspect, the ratio of the fuel injection amount to the total fuel amount used for combustion becomes a parameter indicating the stable adhesion state of the wall-attached fuel, and the smooth value obtained by smoothing it in the time direction is: This is a parameter indicating the current adhesion state of the fuel adhering to the wall. Therefore, by acquiring these deviations as parameters related to the stability of the fuel adhesion state, it is possible to accurately estimate the stability of the fuel adhesion state.

  According to the fourteenth aspect of the invention, the lower the coolant temperature of the internal combustion engine, the more unstable the fuel adhesion state. By acquiring the coolant temperature as a parameter related to the stability of the fuel adhesion state, the fuel adhesion state is stabilized. It is possible to accurately estimate the degree.

  According to the fifteenth aspect, the shorter the elapsed time after returning from the fuel cut, the more fuel adheres to the wall surface, so by acquiring the elapsed time as a parameter related to the stability of the fuel adhesion state, It becomes possible to accurately estimate the stability of the fuel adhesion state.

  According to the sixteenth aspect of the invention, the deviation between the increase coefficient reflected in the basic fuel amount for increasing the fuel injection amount and the smooth value in the time direction indicates how much the effect of the fuel increase is after the return from the fuel increase. It becomes a parameter indicating whether it remains. As the effect of fuel increase remains, the fuel desorbs from the wall surface. Therefore, by obtaining the deviation as a parameter related to the stability of the fuel adhesion state, the stability of the fuel adhesion state is accurately estimated. It becomes possible.

Embodiment 1 FIG.
The first embodiment of the present invention will be described below with reference to FIGS.

[Description of engine system configuration]
FIG. 1 is a schematic configuration diagram of an engine system to which an air-fuel ratio control apparatus for an internal combustion engine as Embodiment 1 of the present invention is applied. An intake passage 6 and an exhaust passage 8 are connected to the combustion chamber 4 of the internal combustion engine 2 according to the present embodiment. A throttle 14 for adjusting the flow rate of air flowing into the combustion chamber 4 is disposed in the intake passage 6. A fuel injection valve 12 for supplying fuel to the combustion chamber 4 is attached in the vicinity of the combustion chamber 4 in the intake passage 6.

  The fuel injected from the fuel injection valve 12 is supplied from the fuel tank 20 through a fuel passage (not shown). A vapor passage 24 is connected to the fuel tank 20 for extracting evaporated fuel generated inside. One end of the vapor passage 24 is connected to a canister 18 filled with activated carbon for adsorbing evaporated fuel. Therefore, the evaporated fuel generated inside the fuel tank 20 reaches the canister 18 through the vapor passage 24 and is adsorbed inside the canister 18. Connected to the canister 18 is an air supply passage 26 having one end opened to the atmosphere, and a purge passage 22 communicating with the intake passage 6 downstream of the throttle valve 14. The purge passage 22 is provided with a purge control valve 16 for controlling the flow rate of gas flowing through the purge passage 22. The purge control valve 16 is an electromagnetic valve that realizes an arbitrary opening degree by duty control. The purge control valve 16 corresponds to a purge mechanism according to the present invention.

  The internal combustion engine 2 includes an ECU (Electronic Control Unit) 40 as its control device. The ECU 40 comprehensively controls various devices related to the operation of the internal combustion engine 2 based on the operation data of the internal combustion engine 2 detected by a plurality of sensors. A fuel injection valve 12 and a purge control valve 16 are connected to the output side of the ECU 40.

  A water temperature sensor 30, a rotation speed sensor 32, an exhaust gas sensor 34, an air flow meter 36, and an accelerator position sensor 38 are connected to the input side of the ECU 40. The water temperature sensor 30 is attached to the water jacket of the internal combustion engine 2 and outputs a signal corresponding to the cooling water temperature. The cooling water temperature represents the engine temperature of the internal combustion engine 2. The rotational speed sensor 32 is attached in the vicinity of the crankshaft and outputs a signal corresponding to the engine rotational speed. The exhaust gas sensor 34 is provided in the exhaust passage 8 and outputs a signal corresponding to the air-fuel ratio (exhaust air-fuel ratio) of the exhaust gas. The air flow meter 36 is disposed upstream of the throttle 14 and outputs a signal corresponding to the intake air flow rate. The accelerator position sensor 38 is attached to an accelerator pedal (not shown) and outputs a signal corresponding to the accelerator position.

  The ECU 40 receives operation data from the sensors 30, 32, 34, 36, and 38 and supplies drive signals to the devices 12 and 16. In addition to the sensors 32, 34, 36, and 38 and the devices 12 and 16, a plurality of sensors and devices are connected to the ECU 40, but the description thereof is omitted here.

[Overview of purge control]
One of the controls of the internal combustion engine 2 performed by the ECU 40 is purge control for purging the evaporated fuel adsorbed by the canister 18. This purge control is executed by appropriately driving the purge control valve 16 when the predetermined purge execution condition is satisfied during operation of the internal combustion engine 2. The purge execution condition is that the operating state of the internal combustion engine 2 determined from the engine speed and the intake air flow rate is within a predetermined purge execution region.

  The duty output to the purge control valve 16 is controlled according to the target value of the purge rate. The purge rate is the ratio of the purge gas flow rate to the intake air flow rate, and can be controlled by the duty of the purge control valve 16. The target value of the purge rate is stored in the map in association with the engine speed and the intake air flow rate. At the time of purge control, the target value of the purge rate corresponding to the current engine speed and intake air flow rate is read from this map, and the purge control valve 16 is duty driven according to the value.

  When the purge control valve 16 is driven with a duty, the negative pressure in the intake passage 6 of the internal combustion engine 2 is guided to the canister 18. As a result, the evaporated fuel in the canister 18 is discharged into the purge passage 22 as purge gas together with the air sucked from the atmospheric supply passage 26. The discharged purge gas is supplied to the intake passage 6 through the purge passage 22 and is combusted in the combustion chamber 4.

[Description of air-fuel ratio control]
Further, as one of the controls of the internal combustion engine 2, the ECU 40 performs air-fuel ratio control for setting the air-fuel ratio of the air-fuel mixture in the combustion chamber 4 to a desired target air-fuel ratio. The air-fuel ratio control is to control the fuel injection amount from the fuel injection valve 12 so that the exhaust air-fuel ratio measured by the exhaust gas sensor 34 becomes the target air-fuel ratio during the operation of the internal combustion engine 2. In the system of the present embodiment, when the purge control valve 16 is operated, the evaporated fuel contained in the purge gas in addition to the fuel from the fuel injection valve 12 is also supplied into the combustion chamber 4. Therefore, it is necessary to control the fuel injection amount in consideration of the fluctuations.

The fuel injection amount from the fuel injection valve 12 is determined by the fuel injection time TAU that is the valve opening time of the fuel injection valve 12. The following formula (1) is a formula for calculating the fuel injection time TAU. In the equation (1), TP is a basic injection time, and is calculated by multiplying a ratio (GA / NE) between the engine rotational speed NE and the intake air flow rate GA by a predetermined injection coefficient. KRICH is an increase coefficient set in response to a request to increase the fuel, such as an increase correction during cold, an increase correction during WOT, or an increase correction for protecting the catalyst after fuel cut.
TAU = TP × (KRICH + FAF + FPG) + FMW (1)

  In the above equation (1), FAF is a feedback correction coefficient. The feedback correction coefficient FAF is updated in a decreasing direction in small steps while it is determined from the output of the exhaust gas sensor 34 that the exhaust air-fuel ratio is richer than the target air-fuel ratio. As a result, the fuel injection time TAU gradually decreases, and the exhaust air / fuel ratio measured from the output of the exhaust gas sensor 34 is eventually reversed to be leaner than the target air / fuel ratio. When the exhaust air-fuel ratio reverses from rich to lean than the target air-fuel ratio, the feedback correction coefficient FAF is greatly skipped in the increasing direction at that time. Then, the feedback correction coefficient FAF is updated in an increasing direction in small steps until the exhaust air-fuel ratio reverses richer than the target air-fuel ratio. As a result, the fuel injection time TAU increases little by little, and the exhaust air-fuel ratio eventually reverses from lean to richer than the target air-fuel ratio. When the exhaust air-fuel ratio reverses from lean to richer than the target air-fuel ratio, the feedback correction coefficient FAF is largely skipped in the decreasing direction at that time. Subsequently, the exhaust air / fuel ratio is maintained in the vicinity of the target air / fuel ratio by repeatedly executing the updating process described above.

  However, when the purge gas is purged from the canister 18 to the intake passage 6, the influence causes a change in the air-fuel ratio of the air-fuel mixture in the combustion chamber 4. If the air-fuel ratio of the purge gas is equal to the target air-fuel ratio, the air-fuel ratio of the air-fuel mixture can be made the target air-fuel ratio by setting the ratio of the fuel injection amount to the intake air flow rate to the target air-fuel ratio. However, if the purge gas is rich, the air-fuel ratio of the air-fuel mixture cannot be made the target air-fuel ratio unless the fuel injection amount is reduced to such an extent that the influence can be offset. Similarly, if the purge gas is lean, the air-fuel mixture cannot be brought to the target air-fuel ratio unless the fuel injection amount is increased appropriately.

Therefore, in the above equation (1), the purge correction coefficient FPG is set as a correction coefficient to be applied to the fuel injection time TAU in order to cancel the influence of the purge gas. The purge correction coefficient FPG is calculated as the product of the vapor concentration correction coefficient FGPG and the purge rate PGR as shown in the following equation (2).
FPG = FGPG × PGR (2)

  The vapor concentration correction coefficient FGPG in the equation (2) is a correction coefficient corresponding to the fuel concentration of the purge gas (the concentration of the evaporated fuel in the purge gas), and is a correction to be applied to the fuel injection time TAU for the purge rate PGR of 1%. It has a meaning as a quantity. This correction amount needs to be a larger negative value as the fuel concentration of the purge gas is higher (richer), and to a larger positive value as the fuel concentration of the purge gas is lower (lean). There is.

  Since the influence of the supply of the purge gas on the air-fuel ratio varies depending on the fuel concentration of the purge gas, it is important to know the fuel concentration of the purge gas in order to maintain the air-fuel ratio close to the target air-fuel ratio with high accuracy. The ECU 40 learns the vapor concentration correction coefficient FGPG corresponding to the fuel concentration of the purge gas from the steady deviation from the reference value of the feedback correction coefficient FAF during execution of the purge control. A fuel concentration learning method, that is, a vapor concentration correction coefficient FGPG learning method will be described later.

  Further, FMW in the above equation (1) is a wall surface adhesion correction amount, which is a correction amount set to cancel the change in the wall surface adhesion fuel amount during the transient operation of the internal combustion engine 2. During transient operation of the internal combustion engine 2, the amount of fuel adhering to the wall surface changes as the intake air flow rate or fuel injection amount increases or decreases. When the amount of fuel adhering to the wall surface increases, the amount of fuel used for combustion becomes smaller than the fuel injection amount, leading to a lean air-fuel ratio. On the other hand, in the process in which the amount of fuel adhering to the wall surface decreases, more fuel than the fuel injection amount is used for combustion, leading to enrichment of the air-fuel ratio. Therefore, in the air-fuel ratio control, the wall injection correction amount FMW is included in the fuel injection time TAU so as to prevent fluctuations in the air-fuel ratio due to changes in the wall surface attached fuel amount. The wall surface adhesion correction amount FMW is stored in the map in association with the engine speed, the intake air flow rate, and the cooling water temperature. At the time of air-fuel ratio control, the wall surface adhesion correction amount FMW corresponding to the current engine speed, intake air flow rate, and cooling water temperature is read from this map and used for calculation of the fuel injection time TAU.

  The adhesion of fuel to the wall surface and the desorption of fuel from the wall surface are also affected by the presence or absence of purge. Since the evaporated fuel is less likely to adhere to the wall surface than the liquid fuel injected from the fuel injection valve 12, the same operating conditions (engine speed, intake air flow rate, This is because the amount of fuel attached to the wall surface at the cooling water temperature) decreases. For this reason, the wall surface adhesion correction amount FMW is corrected according to the presence or absence of the purge or the magnitude of the purge rate.

[Description of basic procedure for fuel concentration learning]
The flowchart of FIG. 2 shows a basic routine used for learning the fuel concentration in the present embodiment. Hereinafter, the basic procedure of density learning according to the present embodiment will be described with reference to the flowchart of FIG. The concentration learning routine shown in the flowchart of FIG. 2 is executed at a predetermined cycle by the ECU 40 that is a control device of the internal combustion engine 2.

In the first step S2 of the routine shown in FIG. 2, the feedback correction coefficient FAF is smoothed according to the following equation (3). In the equation (3), AFF is an air-fuel ratio deviation index value indicating the degree of deviation of the air-fuel ratio from the theoretical air-fuel ratio, and is the feedback correction coefficient FAF or its correction value. NN is a smoothing coefficient and is set to a value of 1 or more. DFPG is a smooth value of AFF, DFPG on the left side is a smooth value after update, and DFPG on the right side is a smooth value before update. The air-fuel ratio deviation index smooth value DFPG is reset when the purge cut is executed, and is maintained at zero during the purge cut.
DFPG = DFPG + (AFF-DFPG) / NN (3)

In the above equation (3), the exhaust gas sensor 34 is a so-called O ₂ sensor, that is, a sensor exhibiting a so-called Z characteristic in which the output suddenly changes between the rich side and the lean side with respect to the exhaust air / fuel ratio. In this case, the feedback correction coefficient FAF is used as it is as the air-fuel ratio deviation index value AFF. On the other hand, when the exhaust gas sensor 34 is a so-called linear air-fuel ratio sensor (A / F sensor), that is, a sensor showing linear output characteristics with respect to the exhaust air-fuel ratio, the feedback correction coefficient FAF is expressed by the following equation (4). Is preferably used as the air-fuel ratio deviation index value AFF. In equation (4), ABYF is an exhaust air-fuel ratio measured by a linear air-fuel ratio sensor, and ABYF0 is a stoichiometric air-fuel ratio. In the present embodiment, it is assumed that a linear air-fuel ratio sensor is provided as the exhaust gas sensor 34. The air-fuel ratio deviation index value AFF becomes a negative value as the actual air-fuel ratio shifts to the rich side from the theoretical air-fuel ratio, and becomes a positive value as the actual air-fuel ratio shifts to the lean side.
AFF = FAF + (ABYF / ABYF0) (4)

  In the next step S4, the value of the learning counter CFGPG is increased by a certain value. In step S6, it is determined whether or not the learning counter CFGPG has reached a predetermined reference number KCS. Until the learning counter CFGPG reaches the reference number KCS, the subsequent processing is skipped and concentration learning is not executed. When the learning counter CFGPG reaches the reference count KCS, the learning counter CFGPG is reset in step S8, and then the processes in and after step S10 are executed. The density learning is executed at a constant learning cycle by the processes of steps S4, S6, and S8. When the purge cut is executed, the learning counter CFGPG is reset at that stage, and is maintained at zero during the purge cut.

  Steps S10 and S12 are processes for determining whether or not the air-fuel ratio deviation exceeds a predetermined dead band. In step S10, it is determined whether the air-fuel ratio deviation index value AFF is greater than or equal to the upper limit value KAFFH of the dead band. If the air-fuel ratio deviation index value AFF is less than the dead band upper limit value KAFFH, the process proceeds to step S12, and it is determined whether or not the air-fuel ratio deviation index value AFF is equal to or smaller than the dead band lower limit value KAFFL. If the air-fuel ratio deviation index value AFF is within the dead band as a result of these determinations, the concentration degree learning is prohibited and the subsequent processing is skipped. On the other hand, when the air-fuel ratio deviation index value AFF exceeds the dead band, the processing after step S14 is executed. By providing the dead zone for the concentration learning in this way, it is possible to prevent the concentration learning from being frequently executed according to the deviation of the air / fuel ratio to the extent of an error, and maintain the stability of the air / fuel ratio control.

  In step S14, it is determined whether or not the purge rate PGR is greater than zero, that is, whether or not purge is currently being executed. The deviation of the air-fuel ratio in the situation where the purge is not executed is due to factors other than the supply of the purge gas, and if the concentration learning is continued in such a situation, the fuel concentration is erroneously learned. Therefore, the concentration learning process after step S16 is executed only when the purge rate PGR is greater than zero as a result of the determination in step S14. When the purge rate PGR is zero, the concentration degree learning is stopped and the subsequent processing is skipped.

  In step S16, it is determined whether or not the purge rate PGR is equal to or higher than a predetermined reference purge rate KPGRL. The reference purge rate KPGRL is set to a value larger than zero and smaller than the maximum purge rate. If the purge rate PGR is greater than or equal to the reference purge rate KPGRL, it is determined in step S18 whether or not concentration learning has been completed. Similarly, if the purge rate PGR is less than the reference purge rate KPGRL, it is determined in step S20 whether concentration learning has been completed. Completion of the density learning can be determined, for example, based on whether or not the vapor density correction coefficient FGPG that is the learning value has converged.

A specific method of concentration learning is to determine the learning update amount DLFGPG of the vapor concentration correction coefficient FGPG according to the air-fuel ratio deviation index smooth value DFPG, as will be described later. Each determination of steps S16, S18, and S20 is a process for selecting an appropriate method according to the purge gas supply state and the progress of concentration learning as a method for determining the learning update amount DLFGPG. If the result of determination is that the purge rate PGR is greater than or equal to the reference purge rate KPGRL and learning has been completed, first, the learning update coefficient KDLFG is determined in step S22. The learning update coefficient KDLFG is set according to the value of the air-fuel ratio deviation index smooth value DFPG, as shown in the graph in the figure. In the next step S30, the learning update amount DLFGPG is calculated by the following equation (5) using the learning update coefficient KDLFG, the purge rate PGR, and the air-fuel ratio deviation index smooth value DFPG.
DLFGPG = DFPG / PGR / 2 × KDLFG (5)

  If the learning update amount DLFGPG is calculated in step S24, the value of the density update count CFGCNG is increased by a fixed value in step S32.

  As a result of the determination in steps S16 and S18, when the purge rate PGR is equal to or greater than the reference purge rate KPGRL and learning is not completed, the learning update amount DLFGPG is determined in step S24. The learning update amount DLFGPG is set according to the value of the air-fuel ratio deviation index smooth value DFPG, as shown in the graph in the figure.

  If the result of determination in steps S16 and S20 is that the purge rate PGR is less than the reference purge rate KPGRL and learning has been completed, the learning update amount DLFGPG is determined in step S26. The learning update amount DLFGPG is set according to the value of the air-fuel ratio deviation index smooth value DFPG, as shown in the graph in the figure.

  If the result of determination in steps S16 and S20 is that the purge rate PGR is less than the reference purge rate KPGRL and learning has not been completed, the learning update amount DLFGPG is determined in step S28. The learning update amount DLFGPG is set according to the value of the air-fuel ratio deviation index smooth value DFPG, as shown in the graph in the figure. As can be seen by comparing the graphs, the learning update amount DLFGPG determined in step S28 is compared with the learning update amount DLFGPG determined in steps S24 and S26 when compared with the same air-fuel ratio deviation index smooth value DFPG. Is also a small value.

  If the learning update amount DLFGPG is determined in any one of steps S30, S24, S26, and S28, the processes in and after step S34 are executed. In step S34, an update guard value KDLFGM for limiting the learning update amount DLFGPG is determined. The update guard value KDLFGM is set according to the value of the air-fuel ratio deviation index smooth value DFPG, as shown in the graph in the figure. In the next step S36, it is determined whether or not the learning update amount DLFGPG is greater than or equal to the update guard value KDLFGM. If the learning update amount DLFGPG is less than the update guard value KDLFGM, the learning update amount DLFGPG is used as it is. On the other hand, if the learning update amount DLFGPG is greater than or equal to the update guard value KDLFGM, the learning update amount DLFGPG is limited to the update guard value KDLFGM in the next step S38. That is, the update guard value KDLFGM is replaced with the learning update amount DLFGPG.

In step S40, the vapor concentration correction coefficient FGPG is updated by the following equation (6) using the learning update amount DLFGPG determined by the above processing. In the equation (6), FGPG on the left side is a vapor concentration correction coefficient after update, and FGPG on the right side is a vapor concentration correction coefficient before update.
FGPG = FGPG + DLFGPG (6)

In the final step S42, the feedback correction coefficient FAF is corrected by the following equation (7), and the air-fuel ratio deviation index smooth value DFPG is corrected by the equation (8). The process of step S42 is a process for preventing the learning update amount DLFGPG from being double corrected between the feedback correction coefficient FAF and the purge correction coefficient FPG. In equation (7), the FAF on the left side is a feedback correction coefficient after correction, and the FAF on the right side is a feedback correction coefficient before correction. In equation (8), the DFPG on the left side is the corrected air-fuel ratio deviation index smooth value, and the DFPG on the right side is the air-fuel ratio deviation index smooth value before correction.
FAF = FAF + DLFGPG × PGR (7)
DFPG = DFPG + DLFGPG × PGR (8)

[Problems associated with changes in the amount of fuel attached to the wall]
As described above, the change in the wall-attached fuel amount during the transient operation of the internal combustion engine 2 is offset by including the wall-attachment correction amount FMW in the fuel injection time TAU. However, the instable state of the fuel adhering to the wall surface is not limited to the transient operation. As a specific example, even when the supply amount of the evaporated fuel changes with purge cut or purge start, the adhesion state of the fuel adhering to the wall surface tends to become unstable. Since the supply of the evaporated fuel affects the air-fuel ratio, the purge correction coefficient FPG is reflected in the basic injection time TP in calculating the fuel injection time TAU as described above. However, if the fuel injection amount of the fuel injection valve 12 changes so as to offset the change in the supply amount of the evaporated fuel, the stable amount of fuel attached to the wall surface of the intake passage (stable wall surface attached fuel amount) also changes. It will be. For this reason, when the supply amount of the evaporated fuel changes, the fuel adheres to the wall surface and the fuel desorbs from the wall surface so as to compensate for the difference in the stable wall surface adhering fuel amount before and after that. A difference between the fuel amount actually used for combustion and the fuel injection amount occurs with a change in the amount of fuel adhering to the wall surface, and a deviation occurs between the target air-fuel ratio and the actual air-fuel ratio.

  According to the above-described air-fuel ratio control, the deviation of the air-fuel ratio accompanying the change in the amount of fuel attached to the wall surface is absorbed by the feedback correction coefficient FAF. However, as described above, the vapor concentration correction coefficient FGPG is learned from the feedback correction coefficient FAF. For this reason, even the deviation of the feedback correction coefficient FAF caused by the change in the amount of fuel attached to the wall surface is reflected in the vapor concentration correction coefficient FGPG. That is, the vapor concentration correction coefficient FGPG is erroneously learned as a result of the change in the fuel concentration of the purge gas, even though the air-fuel ratio has changed due to the change in the amount of fuel attached to the wall surface. The learning value vapor concentration correction coefficient FGPG affects not only the present time but also in the future, so if incorrect learning is performed, the accuracy of air-fuel ratio control will be improved for a while after the wall surface fuel adheres to a stable state. It will be reduced.

[Features of air-fuel ratio control according to the first embodiment]
Therefore, in the air-fuel ratio control according to the present embodiment, regarding the learning of the fuel concentration, based on the concentration learning routine shown in the flowchart of FIG. 2, the learning method correction control is performed as described below. Yes. In the correction control, a parameter related to the stability of the adhesion state of the wall-attached fuel is acquired, and the fuel concentration learning method is corrected according to the acquired parameter value. If the learning method is modified according to the stability of the adhesion state of the fuel adhering to the wall, even if the fuel adhering state becomes unstable as the purge amount increases or decreases, etc. It becomes possible to prevent erroneous learning of concentration.

  In the present embodiment, as a specific method for correcting the learning method, the dead zone of the air-fuel ratio deviation index value AFF according to the determination in steps S10 and S12 of the concentration learning routine is set to a low stability of the fuel adhesion state. The method of enlarging according to this is taken. As long as the air-fuel ratio deviation index value AFF changes within the dead band, the vapor concentration correction coefficient FGPG is not updated, and the air-fuel ratio deviation is not reflected in the vapor concentration correction coefficient FGPG. Therefore, when the stability of the fuel adhesion state is low, the dead zone in which the vapor concentration correction coefficient FGPG is not updated is expanded, so that the influence of the change in the amount of fuel adhering to the wall on the vapor concentration correction coefficient FGPG can be suppressed. .

  In addition, the ratio of the fuel injection amount to the total fuel amount (sum of fuel injection amount and evaporated fuel amount) used for combustion is used as a parameter indicating the adhesion state (stable adhesion state) of the fuel adhering to the wall surface when stable can do. This ratio becomes the maximum value 1 when there is no supply of evaporated fuel, and becomes smaller as the ratio of the fuel injection amount to the total fuel amount becomes smaller. This corresponds to the fact that the amount of fuel adhering to the wall surface decreases when the fuel injection amount decreases with the supply of the evaporated fuel. Hereinafter, the ratio is referred to as a basic adhesion ratio.

  In addition, since the actual adhesion state of the wall-attached fuel changes continuously with a delay in the change in the amount of wall-attached fuel at the time of stabilization, smoothing the basic attachment ratio in the time direction can The change can be estimated with high accuracy. Here, the smooth value of the basic adhesion rate is used as a parameter indicating the current adhesion state of the wall adhesion fuel. Then, the deviation between the basic adhesion ratio and the smooth value of the basic adhesion ratio is acquired as a parameter related to the stability of the fuel adhesion state, and the dead zone is expanded according to the deviation value.

[Specific Processing in Embodiment 1]
In the present embodiment, the dead band is expanded according to the flowchart of FIG. The flowchart of FIG. 3 shows a dead zone expansion control routine as correction control of the learning method. The routine shown in FIG. 3 is executed by the ECU 40, which is a control device for the internal combustion engine 2, together with the routine shown in FIG.

In the first step S100 of the routine shown in FIG. 3, the basic adhesion rate WPBS is calculated by the following equation (9). In the equation (9), TP × KRICH corresponds to the total amount of fuel used for combustion. TP × (KRICH + FPG) corresponds to the fuel injection amount by the fuel injection valve 12. According to the equation (9), the purge correction coefficient FPG becomes a large negative value and the basic adhesion rate WPBS becomes small as the supply amount of the evaporated fuel by the purge increases. As described above, the basic adhesion rate WPBS has a meaning as a parameter indicating the stable adhesion state of the wall-attached fuel.
WPBS = TP × (KRICH + FPG) / (TP × KRICH) (9)

  In the next step S102, it is determined whether or not a fuel cut (F / C) is being executed. Since the basic injection time TP is set to zero during the fuel cut, the value of the basic adhesion rate WPBS is not determined in the equation (9). On the other hand, during the fuel cut, the fuel desorbs from the wall surface, and the fuel adhering to the wall surface is almost lost. Therefore, if it is determined that the fuel cut is being executed, the basic adhesion rate WPBS is set to zero in step S104.

  In the next step S106, a smoothing coefficient n used for smoothing the basic adhesion rate WPBS is determined based on the cooling water temperature THW. The smoothing coefficient n is a value of 1 or more, and is set to a smaller value as the cooling water temperature THW is higher, as shown in the figure. This is because as the cooling water temperature THW increases, that is, as the warm-up of the internal combustion engine 2 progresses, the wall surface temperature of the intake passage 6 increases, and the adhesion state of the wall-attached fuel becomes faster.

In the next step S108, using the smoothing coefficient n determined in step S106, the basic adhesion ratio WPBS is smoothed in the time direction according to the following equation (10). In the equation (10), WP is a smooth value of the basic adhesion ratio WPBS, WP on the left side is the updated smooth value, and WP on the right side is the smooth value before the update. As described above, the basic adhesion rate smoothing value WP has a meaning as a parameter indicating the current adhesion state of the wall-attached fuel.
WP = WP + (WPBS-WP) / n (10)

In the next step S110, as shown in the following equation (11), a deviation ΔWP between the basic adhesion rate smooth value WP calculated in step S108 and the basic adhesion rate WPBS calculated in step S100 or step S104 is calculated. Is done. As described above, the deviation ΔWP has a meaning as a parameter related to the stability of the fuel adhesion state. The larger the deviation ΔWP is, the more positive the amount of fuel adhering to the wall surface is, the more the fuel is desorbed from the wall surface. Conversely, as the deviation ΔWP is a negative value, the amount of fuel attached to the wall surface is insufficient with respect to the stable amount, that is, the fuel adheres to the wall surface.
ΔWP = WP−WPBS (11)

  In the next step S112, the upper limit value KAFFH and the lower limit value KAFFL of the dead band are determined according to the deviation ΔWP calculated in step S110. As shown in the figure, when the deviation ΔWP is within a predetermined range centered on zero, both the upper limit value KAFFH and the lower limit value KAFFL are constant values. However, when the deviation ΔWP exceeds a predetermined range and becomes a large positive value, the lower limit value KAFFL is expanded to a large negative value according to the value of the deviation ΔWP. On the other hand, the upper limit value KAFFH remains a constant value regardless of the value of the deviation ΔWP. That is, in a situation where fuel desorption from the wall surface proceeds, the dead zone is expanded in the negative direction. Conversely, when the deviation ΔWP exceeds a predetermined range and becomes a large negative value, the upper limit value KAFFH is expanded to a large positive value according to the value of the deviation ΔWP. On the other hand, the lower limit value KAFFL remains a constant value regardless of the value of the deviation ΔWP. In other words, in a situation where fuel adheres to the wall surface, the dead zone is expanded in the positive direction.

  The dead band upper limit value KAFFH determined in the routine shown in FIG. 3 is taken into the routine shown in FIG. 2 and used for the determination in step S10. Further, the lower limit value KAFFL of the dead band determined by the routine shown in FIG. 3 is taken into the routine shown in FIG. 2 and used for the determination in step S12.

  In the situation where the fuel adheres to the wall surface as in the purge cut, the air-fuel ratio deviation index value AFF is shifted in the positive direction as the air-fuel ratio becomes leaner. The deviation of the air-fuel ratio deviation index value AFF is reflected in the vapor concentration correction coefficient FGPG. However, according to the routine shown in FIG. 3, the upper limit value KAFFH of the dead band is increased in a situation where fuel adheres to the wall surface. Therefore, even if the air-fuel ratio deviation index value AFF deviates in the positive direction, the upper limit value KAFFH is expanded to fall within the dead band, and the deviation of the air-fuel ratio deviation index value AFF is reflected in the vapor concentration correction coefficient FGPG. This is prevented.

  On the other hand, in the situation where fuel desorption proceeds from the wall surface at the start of the purge, the air-fuel ratio deviation index value AFF is shifted in the negative direction as the air-fuel ratio is enriched. The deviation of the air-fuel ratio deviation index value AFF is reflected in the vapor concentration correction coefficient FGPG. However, according to the routine shown in FIG. 3, the lower limit value KAFFL of the dead band is expanded in the situation where fuel desorption proceeds from the wall surface. . Therefore, even if the air-fuel ratio deviation index value AFF deviates in the negative direction, the lower limit value KAFFL is expanded to fall within the dead band, and the deviation of the air-fuel ratio deviation index value AFF is reflected in the vapor concentration correction coefficient FGPG. This is prevented.

  As described above, by performing the dead zone expansion control as the correction control by executing the routine shown in FIG. 3, the dead zone in which the vapor concentration correction coefficient FGPG is not updated according to the low stability of the fuel adhesion state. And the influence of the change in the amount of fuel adhering to the wall surface on the vapor concentration correction coefficient FGPG can be suppressed. In addition, according to the above-described dead zone expansion control, the dead zone expansion direction is limited to the direction of change in the amount of fuel adhering to the wall surface, so that the vapor concentration correction is performed as usual in the direction not affected by the change in the amount of fuel adhering to the wall surface. The coefficient FGPG can be updated, and a decrease in learning accuracy can be suppressed.

  In the present embodiment, the “fuel injection amount calculating means” according to the first aspect of the present invention is realized by the ECU 40 calculating the fuel injection time TAU according to the equation (1) in the air-fuel ratio control described above. . The fuel injection time TAU calculated by the equation (1) corresponds to the fuel injection amount calculated by the “fuel injection amount calculating means”. In the course of calculating the fuel injection time TAU, the ECU 40 calculates the feedback correction coefficient FAF, thereby realizing the “feedback correction coefficient calculating means” according to the first aspect of the invention. Further, in the process of calculating the fuel injection time TAU, the “purge correction coefficient calculation means” according to the first aspect of the present invention is realized by the ECU 40 calculating the purge correction coefficient FPG.

  Further, in the present embodiment, the “concentration learning means” according to the first and second inventions is realized by the ECU 40 executing the routine shown in FIG. The vapor concentration correction coefficient FGPG calculated by the routine shown in FIG. 2 corresponds to the fuel concentration of the purge gas learned by the “concentration learning means”.

  Further, in the present embodiment, the “learning method correcting means” according to the first, second, third and thirteenth inventions is realized by the ECU 40 executing the routine shown in FIG. The deviation ΔWP calculated by the routine shown in FIG. 3 is a parameter related to the stability of the adhesion state of the wall-attached fuel obtained by the “learning method correcting means” in the first, second, third and thirteenth inventions. It corresponds to.

Embodiment 2. FIG.
Next, a second embodiment of the present invention will be described with reference to FIG. The system of the present embodiment is realized by causing the ECU 40 to execute a routine shown in FIG. 4 described later in place of the routine shown in FIG. 3 described above in the system of the first embodiment.

[Features of air-fuel ratio control according to the second embodiment]
Also in the air-fuel ratio control according to the present embodiment, regarding the learning of the fuel concentration, correction control of the learning method is further performed as described below based on the concentration learning routine shown in the flowchart of FIG. The correction control according to the present embodiment is similar to the correction control according to the first embodiment. The method of enlarging according to the low of the.

  In the correction control according to the present embodiment, the cooling water temperature of the internal combustion engine 2 is acquired as a parameter related to the stability of the fuel adhesion state, and the dead zone is expanded according to the low cooling water temperature. The adhesion state of the wall-attached fuel is particularly unstable when cold, and becomes stable as the internal combustion engine warms up. Therefore, it is considered that mislearning during cold can be effectively prevented if the dead zone is expanded according to the low coolant temperature.

[Specific Processing in Second Embodiment]
In the present embodiment, the dead band is expanded according to the flowchart of FIG. The flowchart of FIG. 4 shows a dead zone expansion control routine as correction control of the learning method. The routine shown in FIG. 4 is executed by the ECU 40, which is a control device for the internal combustion engine 2, together with the routine shown in FIG.

  In the first step S120 of the routine shown in FIG. 4, the dead zone expansion value KEXP is determined based on the coolant temperature THW. As shown in the figure, the enlarged value KEXP is set to a smaller value as the cooling water temperature THW becomes higher, and is set to zero above the cooling water temperature THW. This is because as the cooling water temperature THW increases, that is, as the warm-up of the internal combustion engine 2 progresses, the wall surface temperature of the intake passage 6 also increases and the adhesion state of the wall-attached fuel becomes stable.

In the next step S122, the upper limit value KAFFH of the dead band is expanded by the following equation (12) using the expansion value KEXP obtained in step S120. KAFFH on the right side is a reference upper limit value (constant value), and KAFFH on the left side is an enlarged upper limit value.
KAFFH = KAFFH + KEXP (12)

In step S124, the lower limit value KAFFL of the dead band is expanded by the following equation (13) using the expansion value KEXP obtained in step S120. KAFFL on the right side is the reference lower limit value (constant value), and KAFFL on the left side is the expanded lower limit value.
KAFFL = KAFFL−KEXP (13)

  The upper limit value KAFFH of the dead band determined in the routine shown in FIG. 4 is taken into the routine shown in FIG. 2 and used for the determination in step S10. Further, the lower limit value KAFFL of the dead band determined in the routine shown in FIG. 4 is taken into the routine shown in FIG. 2 and used for the determination in step S12.

  During cold weather, the adhesion state of the fuel adhering to the wall surface becomes unstable, and the air-fuel ratio tends to deviate from the target air-fuel ratio due to the fuel adhering to the wall surface and the fuel desorption from the wall surface. In such a situation, the deviation of the air-fuel ratio deviation index value AFF from the reference value increases with the deviation of the air-fuel ratio. The deviation of the air-fuel ratio deviation index value AFF is reflected in the vapor concentration correction coefficient FGPG. However, according to the routine shown in FIG. 4, both the upper limit value KAFFH and the lower limit value KAFFL of the dead band are expanded. Therefore, even if the air-fuel ratio deviation index value AFF deviates in the positive direction or deviates in the negative direction, the upper limit value KAFFH and the lower limit value KAFFL are expanded to fall within the dead band, and the air-fuel ratio deviation The deviation of the index value AFF is prevented from being reflected in the vapor concentration correction coefficient FGPG.

  As described above, the dead band in which the vapor concentration correction coefficient FGPG is not updated according to the low stability of the fuel adhesion state by performing the dead zone expansion control as the correction control by executing the routine shown in FIG. And the influence of the change in the amount of fuel adhering to the wall surface on the vapor concentration correction coefficient FGPG can be suppressed.

  In the present embodiment, the “learning method correcting means” according to the first, second, and fourteenth inventions is realized by the ECU 40 executing the routine shown in FIG.

Embodiment 3 FIG.
Next, a second embodiment of the present invention will be described with reference to FIG. The system of the present embodiment is realized by causing the ECU 40 to execute a routine shown in FIG. 5 described later in place of the routine shown in FIG. 3 described above in the system of the first embodiment.

[Features of air-fuel ratio control according to the third embodiment]
Also in the air-fuel ratio control according to the present embodiment, regarding the learning of the fuel concentration, correction control of the learning method is further performed as described below based on the concentration learning routine shown in the flowchart of FIG. The correction control according to the present embodiment is similar to the correction control according to the first embodiment. The method of enlarging according to the low of the.

  In the correction control according to the present embodiment, as in the second embodiment, the coolant temperature of the internal combustion engine 2 is acquired as a parameter related to the stability of the fuel adhesion state, and the dead zone is expanded according to the low coolant temperature. To do. Further, as in the first embodiment, a deviation ΔWP between the basic adhesion rate WPBS and the smooth value WP of the basic adhesion rate WPBS is also acquired as a parameter related to the stability of the fuel adhesion state, and the deviation ΔWP is also obtained depending on the value of the deviation ΔWP. Expand the dead band. By expanding the dead zone according to the value of deviation ΔWP, it becomes possible to expand the dead zone according to the state of fuel adhering to the wall surface or the state of fuel desorption from the wall surface. It is possible to effectively suppress the influence of the change on the learning value. In addition, by expanding the dead zone according to the low cooling water temperature, it is possible to prevent erroneous learning during cold weather.

[Specific Processing in Embodiment 3]
In the present embodiment, the dead band is expanded according to the flowchart of FIG. The flowchart of FIG. 5 shows a dead zone expansion control routine as correction control of the learning method. The routine shown in FIG. 5 is executed by the ECU 40, which is a control device for the internal combustion engine 2, together with the routine shown in FIG.

  In the first step S130 of the routine shown in FIG. 5, the dead zone expansion coefficient K1 is determined based on the coolant temperature THW. As shown in the figure, the expansion coefficient K1 is set to a smaller value as the cooling water temperature THW increases. This is because as the cooling water temperature THW increases, that is, as the warm-up of the internal combustion engine 2 progresses, the wall surface temperature of the intake passage 6 also increases and the adhesion state of the wall-attached fuel becomes stable.

In the next step S132, the upper limit value KAFFH of the dead band is expanded by the following equation (14) using the expansion coefficient K1 obtained in step S130. KAFFH on the right side is a reference upper limit value (constant value), and KAFFH on the left side is an enlarged upper limit value. ΔWP is a deviation between the basic adhesion rate WPBS and the smooth value WP of the basic adhesion rate WPS, and is calculated by the procedure of steps S100 to S110. However, the deviation ΔWP used in the equation (14) is limited to zero or less. When the deviation ΔWP is larger than zero, zero is substituted into ΔWP. Therefore, the upper limit value KAFFH is increased according to the low coolant temperature THW only in a situation where fuel adheres to the wall surface.
KAFFH = KAFFH × (1−K1 × ΔWP (≦ 0)) (14)

In step S134, the lower limit value KAFFL of the dead band is expanded by the following equation (15) using the expansion coefficient K1 obtained in step S130. KAFFL on the right side is the reference lower limit value (constant value), and KAFFL on the left side is the expanded lower limit value. The deviation ΔWP is calculated according to the steps S100 to S110. The deviation ΔWP used in the equation (15) is limited to zero or more. When the deviation ΔWP is smaller than zero, zero is substituted for ΔWP. Therefore, the lower limit value KAFFL is expanded according to the low coolant temperature THW only in a situation where fuel desorption from the wall surface proceeds.
KAFFL = KAFFL × (1 + K1 × ΔWP (≧ 0)) (15)

  The dead band upper limit value KAFFH determined by the routine shown in FIG. 5 is taken into the routine shown in FIG. 2 and used for the determination in step S10. Further, the lower limit value KAFFL of the dead band determined by the routine shown in FIG. 5 is taken into the routine shown in FIG. 2 and used for the determination in step S12.

  In the situation where the fuel adheres to the wall surface as in the purge cut, the air-fuel ratio deviation index value AFF is shifted in the positive direction as the air-fuel ratio becomes leaner. The deviation of the air-fuel ratio deviation index value AFF is reflected in the vapor concentration correction coefficient FGPG. However, according to the routine shown in FIG. 5, the upper limit value KAFFH of the dead band is increased in a situation where fuel adheres to the wall surface. Further, the adhesion of fuel to the wall surface increases as the cold state in which the adhesion state of the wall surface fuel becomes unstable, but according to the routine shown in FIG. 5, the upper limit value KAFFH increases according to the low coolant temperature. Is done. Therefore, even if the air-fuel ratio deviation index value AFF deviates in the positive direction, the upper limit value KAFFH is expanded to fall within the dead band, and the deviation of the air-fuel ratio deviation index value AFF is reflected in the vapor concentration correction coefficient FGPG. This is prevented.

  On the other hand, in the situation where fuel desorption proceeds from the wall surface at the start of the purge, the air-fuel ratio deviation index value AFF is shifted in the negative direction as the air-fuel ratio is enriched. The deviation of the air-fuel ratio deviation index value AFF is reflected in the vapor concentration correction coefficient FGPG. However, according to the routine shown in FIG. 5, the lower limit value KAFFL of the dead band is increased in the situation where fuel desorption from the wall surface proceeds. . Furthermore, the fuel desorption from the wall surface increases as the cold state in which the adhesion state of the wall surface fuel becomes unstable, but according to the routine shown in FIG. 5, the lower limit value KAFFL is set according to the low cooling water temperature. Enlarged. Therefore, even if the air-fuel ratio deviation index value AFF deviates in the negative direction, the lower limit value KAFFL is expanded to fall within the dead band, and the deviation of the air-fuel ratio deviation index value AFF is reflected in the vapor concentration correction coefficient FGPG. This is prevented.

  As described above, the dead zone in which the vapor concentration correction coefficient FGPG is not updated according to the low stability of the fuel adhesion state by executing the routine shown in FIG. 5 and performing the dead zone expansion control as the correction control. And the influence of the change in the amount of fuel adhering to the wall surface on the vapor concentration correction coefficient FGPG can be suppressed. In addition, according to the above-described dead zone expansion control, the dead zone expansion direction is limited to the direction of change in the amount of fuel adhering to the wall surface. The coefficient FGPG can be updated, and a decrease in learning accuracy can be suppressed.

  In the present embodiment, the “learning method correcting means” according to the first, second, third, thirteenth and fourteenth inventions is realized by the ECU 40 executing the routine shown in FIG. Yes.

Embodiment 4 FIG.
Next, a fourth embodiment of the present invention will be described with reference to FIG. The system of the present embodiment is realized by causing the ECU 40 to execute a routine shown in FIG. 6 described later in place of the routine shown in FIG. 3 described above in the system of the first embodiment.

[Features of air-fuel ratio control according to the fourth embodiment]
Also in the air-fuel ratio control according to the present embodiment, regarding the learning of the fuel concentration, correction control of the learning method is further performed as described below based on the concentration learning routine shown in the flowchart of FIG. The correction control according to the present embodiment is similar to the correction control according to the first embodiment. The method of enlarging according to the low of the.

  In the correction control according to the present embodiment, the elapsed time after returning from the fuel cut is acquired as a parameter related to the stability of the fuel adhesion state, and the dead band is expanded according to the shortness of the elapsed time. When the fuel cut is executed, the fuel is detached from the wall surface, and the fuel attached to the wall surface is almost lost. For this reason, at the time of return from the fuel cut, the fuel adheres to the wall surface, and the air-fuel ratio is easily made lean accordingly. The adhesion of fuel to the wall surface becomes more prominent as the fuel adhering to the wall surface is smaller, that is, the elapsed time after returning from the fuel cut is shorter. Therefore, it is considered that if the dead zone is expanded in accordance with the short elapsed time, erroneous learning at the time of return from the fuel cut can be effectively prevented.

[Specific Processing in Embodiment 4]
In the present embodiment, the dead band is expanded according to the flowchart of FIG. The flowchart in FIG. 6 shows a dead zone expansion control routine as correction control for the learning method. The routine shown in FIG. 6 is executed by the ECU 40, which is a control device for the internal combustion engine 2, together with the routine shown in FIG.

  In the first step S140 of the routine shown in FIG. 6, the dead zone expansion value KEXP is determined based on the elapsed time after returning from the fuel cut. As shown in the figure, the enlarged value KEXP is set to a smaller value as the elapsed time after returning from the fuel cut becomes longer, and is set to zero when the elapsed time is longer than a certain time. This is because the fuel adheres to the wall surface as time elapses after returning from the fuel cut, and the wall surface fuel adheres to a stable state. In addition, the temperature of the internal combustion engine 2 (cooling water temperature) is also taken into account in setting the enlarged value KEXP. As the temperature of the internal combustion engine 2 is higher, the adhesion state of the wall-attached fuel becomes more stable. Therefore, even if the elapsed time after return is the same, the enlarged value KEXP is set to a smaller value if the temperature of the internal combustion engine 2 is higher.

In the next step S142, the upper limit value KAFFH of the dead band is expanded by the following equation (16) using the expansion value KEXP obtained in step S140. KAFFH on the right side is a reference upper limit value (constant value), and KAFFH on the left side is an enlarged upper limit value.
KAFFH = KAFFH + KEXP (16)

  In step S144, the lower limit value KAFFL of the dead band is set to a reference lower limit value (a constant value). That is, the lower limit KAFFL is not increased regardless of the elapsed time after returning from the fuel cut. This is because the phenomenon resulting from the return from the fuel cut is the fuel adhering to the wall surface and does not affect the deviation of the air-fuel ratio deviation index value AFF in the negative direction.

  The dead band upper limit value KAFFH determined in the routine shown in FIG. 6 is taken into the routine shown in FIG. 2 and used for the determination in step S10. Further, the lower limit value KAFFL of the dead band determined by the routine shown in FIG. 6 is taken into the routine shown in FIG. 2 and used for the determination in step S12.

  After returning from the fuel cut, the fuel adheres to the wall surface so as to compensate for the desorption during the fuel cut, so that the air-fuel ratio tends to become lean. In such a situation, the deviation of the air-fuel ratio deviation index value AFF in the positive direction increases as the air-fuel ratio becomes leaner. The deviation of the air-fuel ratio deviation index value AFF is reflected in the vapor concentration correction coefficient FGPG. However, according to the routine shown in FIG. 6, the upper limit value KAFFH of the dead band is expanded according to the elapsed time after returning from the fuel cut. The Therefore, even if the air-fuel ratio deviation index value AFF deviates in the positive direction, the upper limit value KAFFH is expanded to fall within the dead band, and the deviation of the air-fuel ratio deviation index value AFF is reflected in the vapor concentration correction coefficient FGPG. This is prevented.

  As described above, the dead zone in which the vapor concentration correction coefficient FGPG is not updated in accordance with the low stability of the fuel adhesion state by executing the routine shown in FIG. 6 and performing dead zone expansion control as correction control. And the influence of the change in the amount of fuel adhering to the wall surface on the vapor concentration correction coefficient FGPG can be suppressed. In addition, according to the above-described dead zone expansion control, the dead zone expansion direction is limited to the direction in which the return from the fuel cut affects the vapor concentration correction coefficient FGPG as usual in the direction in which the return from the fuel cut does not affect. Can be updated, and a decrease in learning accuracy can be suppressed.

  In the present embodiment, the “learning method correcting means” according to the first, second, third, fourteenth and fifteenth inventions is realized by the ECU 40 executing the routine shown in FIG. Yes.

Embodiment 5. FIG.
Next, a fifth embodiment of the present invention will be described with reference to FIG. The system of the present embodiment is realized by causing the ECU 40 to execute a routine shown in FIG. 7 described later in place of the routine shown in FIG. 3 described above in the system of the first embodiment.

[Features of air-fuel ratio control according to the fifth embodiment]
Also in the air-fuel ratio control according to the present embodiment, regarding the learning of the fuel concentration, correction control of the learning method is further performed as described below based on the concentration learning routine shown in the flowchart of FIG. The correction control according to the present embodiment is similar to the correction control according to the first embodiment. The method of enlarging according to the low of the.

  In the correction control according to the present embodiment, the residual degree of the influence of the fuel increase is acquired as a parameter related to the stability of the fuel adhesion state, and the dead band is expanded according to the magnitude of the residual degree. In the internal combustion engine 2, the fuel injection amount is temporarily increased during WOT. Further, there is a case where the fuel injection amount is temporarily increased to protect the catalyst even after returning from the fuel cut. As the fuel injection amount is increased, the amount of fuel adhering to the wall surface increases, and the amount of fuel adhering to the wall surface temporarily increases. However, after that, when returning from the fuel increase to the normal air-fuel ratio feedback control, the excessively adhered fuel is desorbed from the wall surface, and the air-fuel ratio is easily enriched accordingly. The fuel desorption from the wall surface becomes more prominent as the amount of fuel adhering to the wall surface is excessive, that is, as the influence of the fuel increase remains. Therefore, if the dead zone is expanded according to the residual degree of the influence of the fuel increase, it is considered that erroneous learning at the time of return from the fuel increase can be effectively prevented.

[Specific Processing in Embodiment 5]
In the present embodiment, the dead band is expanded according to the flowchart of FIG. The flowchart of FIG. 7 shows a dead zone expansion control routine as correction control of the learning method. The routine shown in FIG. 7 is executed by the ECU 40, which is a control device for the internal combustion engine 2, together with the routine shown in FIG.

In the first step S150 of the routine shown in FIG. 7, first, the increase coefficient KRICHX is smoothed in the time direction by the following equation (17). The increase coefficient KRICHX is obtained by removing the increase correction amount during the cold time from the increase coefficient KRICH used for calculating the fuel injection time TAU in the equation (1). That is, the increase coefficient KRICHX corresponds to the increase correction at the time of WOT and the increase correction for protecting the catalyst after the fuel cut. In Equation (17), RICHSM is a smooth value of the basic increase coefficient KRICHX, RICHSM on the left side is a smoothed value after update, and RICHSM on the right side is a smooth value before update. n is a smoothing coefficient.
RICHSM = RICHSM + (KRICHX−RICHSM) / n (17)

Subsequently, as shown in the following equation (18), a deviation ΔRICH between the increase coefficient smoothing value RICHSM and the current increase coefficient KRICHX is calculated. The increase coefficient smoothing value RICHSM is a parameter indicating the current amount of fuel adhering to the wall surface, and the current increase coefficient KRICHX is a parameter indicating the stable amount of the fuel amount adhering to the wall surface. Therefore, it can be said that as the deviation ΔRICH is larger, the amount of fuel attached to the wall surface becomes excessive relative to the stable amount, that is, the influence of the fuel increase remains. Therefore, the deviation ΔRICH has a meaning as a parameter related to the stability of the fuel adhesion state.
ΔRICH = RICHSM−KRICHX (18)

  In the next step S152, the dead zone expansion value KEXP is determined based on the deviation ΔRICH calculated in step S150. As shown in the figure, the enlarged value KEXP is set to a large value in proportion to the deviation ΔRICH. This is because as the deviation ΔRICH is larger, the amount of fuel attached to the wall surface becomes excessive with respect to the stable amount, and fuel desorption from the wall surface proceeds. In addition, the temperature of the internal combustion engine 2 (cooling water temperature) is also taken into account in setting the enlarged value KEXP. As the temperature of the internal combustion engine 2 increases, the adhesion state of the wall-attached fuel becomes more stable. Therefore, even if the deviation ΔRICH is the same, the enlarged value KEXP is set to a smaller value if the temperature of the internal combustion engine 2 is higher.

  In step S154, the upper limit value KAFFH of the dead band is set to the reference upper limit value (a constant value). That is, the upper limit value KAFFH is not enlarged regardless of the deviation ΔRICH. This is because the phenomenon resulting from the return from the fuel increase is the detachment of the fuel from the wall surface and does not affect the deviation of the air-fuel ratio deviation index value AFF in the positive direction.

In the next step S156, the lower limit value KAFFL of the dead band is expanded by the following equation (19) using the expanded value KEXP obtained in step S152. KAFFL on the right side is the reference lower limit value (constant value), and KAFFL on the left side is the expanded lower limit value.
KAFFL = KAFFL−KEXP (19)

  The upper limit value KAFFH of the dead band determined in the routine shown in FIG. 7 is taken into the routine shown in FIG. 2 and used for the determination in step S10. Further, the lower limit value KAFFL of the dead band determined in the routine shown in FIG. 7 is taken into the routine shown in FIG. 2 and used for the determination in step S12.

  After returning from the fuel increase, the fuel adsorbs excessively during the fuel increase, so that the fuel desorbs from the wall surface, so that the air-fuel ratio tends to become rich. In such a situation, the deviation of the air-fuel ratio deviation index value AFF in the negative direction increases with the enrichment of the air-fuel ratio. The deviation of the air-fuel ratio deviation index value AFF is reflected in the vapor concentration correction coefficient FGPG. However, according to the routine shown in FIG. 7, the lower limit value KAFFL of the dead band is expanded according to the residual degree of the influence of the fuel increase. Therefore, even if the air-fuel ratio deviation index value AFF deviates in the negative direction, the lower limit value KAFFL is expanded to fall within the dead band, and the deviation of the air-fuel ratio deviation index value AFF is reflected in the vapor concentration correction coefficient FGPG. This is prevented.

  As described above, the dead band in which the vapor concentration correction coefficient FGPG is not updated according to the low stability of the fuel adhesion state by executing the dead zone expansion control as the correction control by executing the routine shown in FIG. And the influence of the change in the amount of fuel adhering to the wall surface on the vapor concentration correction coefficient FGPG can be suppressed. In addition, according to the above-described dead zone expansion control, the dead zone expansion direction is limited to the direction in which the return from the fuel increase affects, so in the direction in which the return from the fuel increase does not affect, the vapor concentration correction coefficient FGPG as usual. Can be updated, and a decrease in learning accuracy can be suppressed.

  In the present embodiment, the “learning method correcting means” according to the first, second, third, fourteenth and sixteenth inventions is realized by the ECU 40 executing the routine shown in FIG. Yes.

Embodiment 6 FIG.
Next, a sixth embodiment of the present invention will be described with reference to FIG. The system of the present embodiment is realized by causing the ECU 40 to execute a routine shown in FIG. 8 described later in place of the routine shown in FIG. 3 described above in the system of the first embodiment.

[Features of air-fuel ratio control according to the sixth embodiment]
Also in the air-fuel ratio control according to the present embodiment, regarding the learning of the fuel concentration, correction control of the learning method is further performed as described below based on the concentration learning routine shown in the flowchart of FIG. The correction control according to the present embodiment differs from the correction control according to the first embodiment in that the learning update amount DLFGPG used for updating the vapor concentration correction coefficient FGPG in the processing of step S40 of the concentration learning routine is changed to the fuel adhesion state. The method of reducing according to the low stability of is taken. According to this, even if a deviation occurs in the air-fuel ratio due to a change in the amount of fuel attached to the wall surface, the amount reflected on the vapor concentration correction coefficient FGPG can be reduced by reducing the learning update amount DLFGPG. In addition, it is possible to suppress the influence of the change in the amount of fuel adhering to the wall surface on the vapor concentration correction coefficient FGPG.

  Further, the correction control according to the present embodiment employs a method of increasing the smoothing coefficient NN related to the smoothing process in step S2 of the concentration learning routine according to the low stability of the fuel adhesion state. According to this, it is possible to reduce the speed at which the change in the amount of fuel adhering to the wall surface is reflected in the air-fuel ratio deviation index smooth value DFPG used for the calculation of the learning update amount DLFGPG. The effect on the vapor concentration correction coefficient FGPG can be suppressed.

  In the correction control according to the present embodiment, the deviation ΔWP between the basic adhesion rate WPBS and the smooth value WP of the basic adhesion rate WPBS is acquired as a parameter related to the stability of the fuel adhesion state, as in the first embodiment. Then, the learning update amount DLFGPG is reduced according to the value of the deviation ΔWP, and the smoothing coefficient NN is increased according to the value of the deviation ΔWP. By performing correction control using the deviation ΔWP as the parameter, the learning update amount DLFGPG and the smoothing coefficient NN can be appropriately changed according to the state of fuel adhesion to the wall surface or the state of fuel desorption from the wall surface. It is possible to effectively suppress the influence of the change in the amount of fuel attached to the wall surface on the learning value.

[Specific Processing in Embodiment 6]
In the present embodiment, the update amount is reduced according to the flowchart of FIG. The flowchart of FIG. 8 shows a routine of update reduction control as correction control of the learning method. The routine shown in FIG. 8 is executed by the ECU 40, which is a control device for the internal combustion engine 2, together with the routine shown in FIG.

  In the first step S200 of the routine shown in FIG. 8, the basic adhesion rate WPBS is calculated by the above equation (9). The basic adhesion rate WPBS has a meaning as a parameter indicating the stable adhesion state of the wall-attached fuel.

  In the next step S202, it is determined whether or not a fuel cut is being executed. If it is determined that the fuel cut is being executed, the basic adhesion rate WPBS is set to zero in step S204.

  In the next step S206, a smoothing coefficient n used for smoothing the basic adhesion rate WPBS is determined based on the cooling water temperature THW. The smoothing coefficient n is a value of 1 or more, and is set to a smaller value as the cooling water temperature THW is higher, as shown in the figure.

  In the next step S208, using the smoothing coefficient n determined in step S206, the basic adhesion rate WPBS is smoothed in the time direction according to the above equation (10). The basic adhesion rate smooth value WP calculated here has a meaning as a parameter indicating the current adhesion state of the wall-attached fuel.

  In the next step S210, as shown in the above equation (11), a deviation ΔWP between the basic adhesion rate smooth value WP calculated in step S208 and the basic adhesion rate WPBS calculated in step S200 or step S204 is calculated. Is done. This deviation ΔWP has a meaning as a parameter related to the stability of the fuel adhesion state.

  In the next step S212, the update ratio KHWP of the learning update amount DLFGPG is determined according to the deviation ΔWP calculated in step S210. As shown in the figure, when the deviation ΔWP is in a predetermined range centered on zero, the update ratio KHWP is set to 1. That is, if the deviation ΔWP is within the predetermined range, the learning update amount DLFGPG calculated by the routine shown in FIG. 2 is used as it is for updating the vapor concentration correction coefficient FGPG. However, when the deviation ΔWP exceeds a predetermined range and becomes a positive large value, that is, in a situation where fuel desorption from the wall surface proceeds, the update ratio KHWP is set to a value smaller than 1 according to the value of the deviation ΔWP. The Further, even when the deviation ΔWP exceeds a predetermined range and becomes a large negative value, that is, even when the fuel adheres to the wall surface, the update ratio KHWP becomes a value smaller than 1 according to the value of the deviation ΔWP. Is set. Note that the temperature (cooling water temperature) of the internal combustion engine 2 is also taken into consideration in the setting of the update ratio KHWP. The higher the temperature of the internal combustion engine 2, the more stable the adhesion state of the wall-attached fuel. Therefore, even if the deviation ΔWP is the same, the update rate KHWP is set to a large value if the temperature of the internal combustion engine 2 is high.

In step S214, the learning update amount DLFGPG is corrected by the following equation (20) using the update ratio KHWP obtained in step S212. In Expression (20), DLFGPG on the right side is a learning update amount before correction, and DLFGPG on the left side is a learning update amount after correction.
DLFGPG = DLFGPG × KHWP (20)

  In the next step S216, a smoothing coefficient NN used for smoothing the air-fuel ratio deviation index value AFF is determined according to the deviation ΔWP calculated in step S210. As shown in the figure, when the deviation ΔWP is in a predetermined range centered on zero, the smoothing coefficient NN is set to a constant value (first predetermined value). However, when the deviation ΔWP exceeds a predetermined range and becomes a positive large value, that is, in a situation where fuel desorption from the wall surface proceeds, the smoothing coefficient NN is set to a second predetermined value that is larger than the first predetermined value. The Further, even when the deviation ΔWP exceeds a predetermined range and becomes a negative large value, that is, even when the fuel adheres to the wall surface, the smoothing coefficient NN becomes a second predetermined value larger than the first predetermined value. Is set. It should be noted that the temperature (cooling water temperature) of the internal combustion engine 2 is also taken into consideration in the setting of the smoothing coefficient NN. The higher the temperature of the internal combustion engine 2, the more stable the adhesion state of the wall-attached fuel. Therefore, even if the deviation ΔWP is the same, the second predetermined value is set to a small value if the temperature of the internal combustion engine 2 is high. Further, the higher the temperature of the internal combustion engine 2 is, the larger the deviation ΔWP is switched to the first predetermined value.

  The learning update amount DLFGPG corrected by the routine shown in FIG. 8 is taken into the routine shown in FIG. 2 and used for updating the vapor concentration correction coefficient FGPG in the process of step S40. Further, the smoothing coefficient NN determined by the routine shown in FIG. 8 is taken into the routine shown in FIG. 2 and used for the smoothing process in step S2.

  In the situation where the fuel adheres to the wall surface as in the purge cut, the air-fuel ratio deviation index value AFF is shifted in the positive direction as the air-fuel ratio becomes leaner. On the other hand, in the situation where fuel desorption proceeds from the wall surface at the start of the purge, the air-fuel ratio deviation index value AFF is shifted in the negative direction as the air-fuel ratio is enriched. The deviation of the air-fuel ratio deviation index value AFF is reflected in the vapor concentration correction coefficient FGPG. However, according to the routine shown in FIG. 8, the learning update amount DLFGPG used for updating the vapor concentration correction coefficient FGPG is reduced. Thus, the amount of reflection of the deviation of the air-fuel ratio deviation index value AFF to the vapor concentration correction coefficient FGPG can be reduced, and erroneous learning of the vapor concentration correction coefficient FGPG can be prevented. Further, according to the routine shown in FIG. 8, the smoothing coefficient NN is increased so that the deviation of the air-fuel ratio deviation index value AFF is reflected in the air-fuel ratio deviation index smooth value DFPG used for calculating the learning update amount DLFGPG. Speed can be reduced, and mislearning of the vapor concentration correction coefficient FGPG can be prevented.

  In the present embodiment, the “learning method correcting means” according to the first, fourth, fifth, thirteenth and fourteenth inventions is realized by the ECU 40 executing the routine shown in FIG. Yes.

Embodiment 7 FIG.
Next, a seventh embodiment of the present invention will be described with reference to FIG. The system of the present embodiment is realized by causing the ECU 40 to execute a routine shown in FIG. 9 described later in place of the routine shown in FIG. 3 described above in the system of the first embodiment.

[Features of air-fuel ratio control according to the seventh embodiment]
Also in the air-fuel ratio control according to the present embodiment, regarding the learning of the fuel concentration, correction control of the learning method is further performed as described below based on the concentration learning routine shown in the flowchart of FIG. In the correction control according to the present embodiment, as in the correction control according to the sixth embodiment, the learning update amount DLFGPG used for updating the vapor concentration correction coefficient FGPG in the process of step S40 of the concentration learning routine is set to the fuel adhesion state. A method of reducing according to the low stability is adopted. Further, in the correction control according to the present embodiment, similarly to the correction control according to the sixth embodiment, the smoothing coefficient NN related to the smoothing process in step S2 of the concentration learning routine is set to a low stability of the fuel adhesion state. A method of increasing the size according to the situation is also adopted.

  In the correction control according to the present embodiment, the elapsed time after returning from the fuel cut is acquired as a parameter related to the stability of the fuel adhesion state, as in the fourth embodiment. Then, the learning update amount DLFGPG is reduced according to the short elapsed time after the return, and the smoothing coefficient NN is increased according to the short elapsed time. The adhesion of fuel to the wall becomes more noticeable as the elapsed time after returning from the fuel cut is shorter, so the learning update amount DLFGPG is reduced according to the shorter elapsed time or the elapsed time is shorter If the smoothing coefficient NN is increased, it is considered that erroneous learning at the time of return from the fuel cut can be effectively prevented.

[Specific Processing in Embodiment 7]
In the present embodiment, the update amount is reduced according to the flowchart of FIG. The flowchart of FIG. 9 shows a routine of update reduction control as correction control of the learning method. The routine shown in FIG. 9 is executed by the ECU 40, which is a control device for the internal combustion engine 2, together with the routine shown in FIG.

  In the first step S220 of the routine shown in FIG. 9, the update ratio KHWPC of the learning update amount DLFGPG is determined based on the elapsed time after returning from the fuel cut. As shown in the figure, the update ratio KHWPC is set to a larger value as the elapsed time after returning from the fuel cut becomes longer, and is set to 1 when the elapsed time exceeds a predetermined time. This is because the fuel adheres to the wall surface as time elapses after returning from the fuel cut, and the wall surface fuel adheres to a stable state. In addition, the temperature of the internal combustion engine 2 (cooling water temperature) is also considered in setting the update ratio KHWPC. The higher the temperature of the internal combustion engine 2, the more stable the adhesion state of the wall-attached fuel. Therefore, even if the elapsed time after return is the same, the update ratio KHWPC is set to a larger value if the temperature of the internal combustion engine 2 is higher.

In step S222, the learning update amount DLFGPG is corrected by the following equation (21) using the update ratio KHWPC obtained in step S220. In equation (21), DLFGPG on the right side is the learning update amount before correction, and DLFGPG on the left side is the learning update amount after correction.
DLFGPG = DLFGPG × KHWPC (21)

  In the next step S224, the smoothing coefficient NN used for smoothing the air-fuel ratio deviation index value AFF is determined based on the elapsed time after returning from the fuel cut. As shown in the figure, the smoothing coefficient NN is set to a constant value (first predetermined value) when the elapsed time after returning from the fuel cut is a predetermined time or more. However, the smoothing coefficient NN is set to a second predetermined value larger than the first predetermined value until the elapsed time reaches a predetermined time. Further, the temperature (cooling water temperature) of the internal combustion engine 2 is also taken into consideration in the setting of the smoothing coefficient NN. The higher the temperature of the internal combustion engine 2, the more stable the adhesion state of the wall surface fuel. Therefore, even if the elapsed time after the return is the same, the second predetermined value is set to a small value if the temperature of the internal combustion engine 2 is high. Further, the higher the temperature of the internal combustion engine 2 is, the earlier it is switched to the first predetermined value.

  The learning update amount DLFGPG corrected by the routine shown in FIG. 9 is taken into the routine shown in FIG. 2 and used for updating the vapor concentration correction coefficient FGPG in the process of step S40. Further, the smoothing coefficient NN determined by the routine shown in FIG. 9 is taken into the routine shown in FIG. 2 and used for the smoothing process in step S2.

  After returning from the fuel cut, the fuel adheres to the wall surface so as to compensate for the desorption during the fuel cut, so that the air-fuel ratio tends to become lean. In such a situation, the deviation of the air-fuel ratio deviation index value AFF in the positive direction increases as the air-fuel ratio becomes leaner. The deviation of the air-fuel ratio deviation index value AFF is reflected in the vapor concentration correction coefficient FGPG. However, according to the routine shown in FIG. 9, the learning update amount DLFGPG is reduced when the elapsed time after returning from the fuel cut is short. Thus, the reflection amount of the deviation of the air-fuel ratio deviation index value AFF to the vapor concentration correction coefficient FGPG can be reduced, and erroneous learning of the vapor concentration correction coefficient FGPG can be prevented. Further, according to the routine shown in FIG. 9, when the elapsed time after returning from the fuel cut is short, the smoothing coefficient NN is increased so that the deviation of the air-fuel ratio deviation index value AFF is calculated in the learning update amount DLFGPG. The speed reflected on the used air-fuel ratio deviation index smooth value DFPG can be reduced, and erroneous learning of the vapor concentration correction coefficient FGPG can be prevented.

  In the present embodiment, the “learning method correcting means” according to the first, fourth, fifth, fourteenth and fifteenth inventions is realized by the ECU 40 executing the routine shown in FIG. Yes.

Embodiment 8 FIG.
Next, an eighth embodiment of the present invention will be described with reference to FIG. The system of the present embodiment is realized by causing the ECU 40 to execute a routine shown in FIG. 10 described later in place of the routine shown in FIG. 3 described above in the system of the first embodiment.

[Features of air-fuel ratio control according to the eighth embodiment]
Also in the air-fuel ratio control according to the present embodiment, regarding the learning of the fuel concentration, correction control of the learning method is further performed as described below based on the concentration learning routine shown in the flowchart of FIG. In the correction control according to the present embodiment, as in the correction control according to the sixth embodiment, the learning update amount DLFGPG used for updating the vapor concentration correction coefficient FGPG in the process of step S40 of the concentration learning routine is set to the fuel adhesion state. A method of reducing according to the low stability is adopted. Further, in the correction control according to the present embodiment, similarly to the correction control according to the sixth embodiment, the smoothing coefficient NN related to the smoothing process in step S2 of the concentration learning routine is set to a low stability of the fuel adhesion state. A method of increasing the size according to the situation is also adopted.

  In the correction control according to the present embodiment, as in the fifth embodiment, the residual degree of the influence of the fuel increase is acquired as a parameter related to the stability of the fuel adhesion state. Then, the learning update amount DLFGPG is reduced according to the magnitude of the residual degree, and the smoothing coefficient NN is increased according to the residual degree. Desorption of fuel from the wall becomes more noticeable as the amount of fuel adhering to the wall is excessive, that is, as the effect of fuel increase remains, so the learning update amount DLFGPG is reduced according to the residual level of the effect of fuel increase. If the smoothing coefficient NN is increased according to the residual degree of the influence of the fuel increase, it is considered that erroneous learning at the time of return from the fuel increase can be effectively prevented.

[Specific Processing in Eighth Embodiment]
In the present embodiment, the update amount is reduced according to the flowchart of FIG. The flowchart of FIG. 10 shows a routine of update reduction control as correction control of the learning method. The routine shown in FIG. 10 is executed by the ECU 40, which is a control device for the internal combustion engine 2, together with the routine shown in FIG.

  In the first step S230 of the routine shown in FIG. 10, first, the increase coefficient KRICHX is smoothed in the time direction according to the above equation (17). Further, the deviation ΔRICH between the increase coefficient smoothing value RICHSM and the current increase coefficient KRICHX is calculated by the above equation (18).

  In the next step S232, the update ratio KHWPR of the learning update amount DLFGPG is determined based on the deviation ΔRICH calculated in step S230. As shown in the figure, the update ratio KHWPR is set to a smaller value as the deviation ΔRICH increases. This is because as the deviation ΔRICH is larger, the amount of fuel attached to the wall surface becomes excessive with respect to the stable amount, and fuel desorption from the wall surface proceeds. In addition, the temperature of the internal combustion engine 2 (cooling water temperature) is also considered in setting the update ratio KHWPR. The higher the temperature of the internal combustion engine 2, the more stable the adhesion state of the wall-attached fuel. Therefore, even if the deviation ΔRICH is the same, the update ratio KHWPR is set to a large value if the temperature of the internal combustion engine 2 is high.

In step S234, the learning update amount DLFGPG is corrected by the following equation (22) using the update ratio KHWPR obtained in step S232. In Expression (22), DLFGPG on the right side is a learning update amount before correction, and DLFGPG on the left side is a learning update amount after correction.
DLFGPG = DLFGPG × KHWPR (22)

  In the next step S236, a smoothing coefficient NN used for smoothing the air-fuel ratio deviation index value AFF is determined based on the deviation ΔRICH calculated in step S230. As shown in the figure, the smoothing coefficient NN is set to a constant value (first predetermined value) when the deviation ΔRICH is equal to or smaller than a predetermined value. However, when the deviation ΔRICH becomes equal to or greater than a predetermined value, the smoothing coefficient NN is set to a second predetermined value that is larger than the first predetermined value. Further, the temperature (cooling water temperature) of the internal combustion engine 2 is also taken into consideration in the setting of the smoothing coefficient NN. The higher the temperature of the internal combustion engine 2, the more stable the adhesion state of the wall-attached fuel. Therefore, even if the deviation ΔRICH is the same, the second predetermined value is set to a small value if the temperature of the internal combustion engine 2 is high. Further, the temperature is switched to the first predetermined value when the deviation ΔRICH is larger as the temperature of the internal combustion engine 2 is higher.

  The learning update amount DLFGPG corrected by the routine shown in FIG. 10 is taken into the routine shown in FIG. 2 and used for updating the vapor concentration correction coefficient FGPG in the process of step S40. Further, the smoothing coefficient NN determined by the routine shown in FIG. 10 is taken into the routine shown in FIG. 2 and used for the smoothing process in step S2.

  After returning from the fuel increase, the fuel adsorbs excessively during the fuel increase, so that the fuel desorbs from the wall surface, so that the air-fuel ratio tends to become rich. In such a situation, the deviation of the air-fuel ratio deviation index value AFF in the negative direction increases with the enrichment of the air-fuel ratio. The deviation of the air-fuel ratio deviation index value AFF is reflected in the vapor concentration correction coefficient FGPG. However, according to the routine shown in FIG. 10, when the residual effect of the fuel increase is large, the learning update amount DLFGPG is reduced, A reflection amount of the deviation of the fuel ratio deviation index value AFF to the vapor concentration correction coefficient FGPG can be reduced, and erroneous learning of the vapor concentration correction coefficient FGPG can be prevented. Further, according to the routine shown in FIG. 10, the smoothing coefficient NN is increased when the influence of the fuel increase is large, so that the deviation of the air-fuel ratio deviation index value AFF is used for calculating the learning update amount DLFGPG. The speed reflected in the air-fuel ratio deviation index smooth value DFPG can be reduced, and erroneous learning of the vapor concentration correction coefficient FGPG can be prevented.

  In the present embodiment, the “learning method correcting means” according to the first, fourth, fifth, fourteenth and sixteenth inventions is realized by the ECU 40 executing the routine shown in FIG. Yes.

Embodiment 9 FIG.
Next, Embodiment 9 of the present invention will be described with reference to FIG. The system of the present embodiment is realized by causing the ECU 40 to execute a routine shown in FIG. 11 described later in place of the routine shown in FIG. 3 in the system of the first embodiment.

[Features of air-fuel ratio control according to the ninth embodiment]
Also in the air-fuel ratio control according to the present embodiment, regarding the learning of the fuel concentration, correction control of the learning method is further performed as described below based on the concentration learning routine shown in the flowchart of FIG. In the correction control according to the present embodiment, unlike the correction control according to the first embodiment, the reference number KCS for the determination in step S6 of the concentration learning routine is increased according to the low stability of the fuel adhesion state. Take the way. By increasing the reference number KCS, the update period of the vapor concentration correction coefficient FGPG can be extended. If the renewal period is extended, the duration of the smoothing process in step S2 is extended accordingly, so that even if there is a temporary deviation in the air-fuel ratio due to the change in the amount of fuel attached to the wall surface The deviation can be sufficiently absorbed by the smoothing process in step S2. As a result, it becomes possible to calculate the learning update amount DLFGPG based on the air-fuel ratio deviation index smooth value DFPG in which the influence of the air-fuel ratio deviation is mitigated, and the influence of the change in the amount of fuel adhering to the wall on the vapor concentration correction coefficient FGPG. Can be suppressed.

  In the correction control according to the present embodiment, the deviation ΔWP between the basic adhesion rate WPBS and the smooth value WP of the basic adhesion rate WPBS is acquired as a parameter related to the stability of the fuel adhesion state, as in the first embodiment. Then, the reference number KCS is increased in accordance with the value of the deviation ΔWP, and the update period of the vapor concentration correction coefficient FGPG is extended. By performing correction control using the deviation ΔWP as the parameter, the update cycle can be appropriately extended according to the state of fuel attachment to the wall surface or the state of fuel desorption from the wall surface, and the amount of fuel attached to the wall surface can be increased. It is possible to effectively suppress the influence of the change on the learning value.

[Specific Processing in Embodiment 9]
In the present embodiment, the update period of the vapor concentration correction coefficient FGPG is extended according to the flowchart of FIG. The flowchart of FIG. 11 shows a routine of update cycle extension control as correction control of the learning method. The routine shown in FIG. 11 is executed by the ECU 40, which is a control device for the internal combustion engine 2, together with the routine shown in FIG.

  In the first step S300 of the routine shown in FIG. 11, the basic adhesion rate WPBS is calculated by the above equation (9). The basic adhesion rate WPBS has a meaning as a parameter indicating the stable adhesion state of the wall-attached fuel.

  In the next step S302, it is determined whether or not a fuel cut is being executed. If it is determined that the fuel cut is being executed, the basic adhesion rate WPBS is set to zero in step S304.

  In the next step S306, the smoothing coefficient n used for smoothing the basic adhesion rate WPBS is determined based on the cooling water temperature THW. The smoothing coefficient n is a value of 1 or more, and is set to a smaller value as the cooling water temperature THW is higher, as shown in the figure.

  In the next step S308, using the smoothing coefficient n determined in step S306, the basic adhesion rate WPBS is smoothed in the time direction according to the above equation (10). The basic adhesion rate smooth value WP calculated here has a meaning as a parameter indicating the current adhesion state of the wall-attached fuel.

  In the next step S310, as shown in the above equation (11), a deviation ΔWP between the basic adhesion rate smooth value WP calculated in step S308 and the basic adhesion rate WPBS calculated in step S300 or step S304 is calculated. Is done. This deviation ΔWP has a meaning as a parameter related to the stability of the fuel adhesion state.

  In step S312, the reference number KCS is determined according to the deviation ΔWP calculated in step S310. As shown in the figure, when the deviation ΔWP is in a predetermined range centered on zero, the reference count KCS is set to a constant value. However, when the deviation ΔWP exceeds a predetermined range and becomes a large positive value, that is, in a situation where fuel desorption from the wall surface proceeds, the reference number KCS is increased according to the value of the deviation ΔWP. Further, even when the deviation ΔWP exceeds a predetermined range and becomes a large negative value, that is, even when fuel adheres to the wall surface, the reference count KCS is increased according to the value of the deviation ΔWP.

  The reference number KCS determined in the routine shown in FIG. 11 is taken into the routine shown in FIG. 2 and used for the determination in step S6.

  In the situation where the fuel adheres to the wall surface as in the purge cut, the air-fuel ratio deviation index value AFF is shifted in the positive direction as the air-fuel ratio becomes leaner. On the other hand, in the situation where fuel desorption proceeds from the wall surface at the start of the purge, the air-fuel ratio deviation index value AFF is shifted in the negative direction as the air-fuel ratio is enriched. The deviation of the air-fuel ratio deviation index value AFF is reflected in the vapor concentration correction coefficient FGPG. However, according to the routine shown in FIG. 8, the reference number KCS is increased and the update period of the vapor concentration correction coefficient FGPG is extended. Thus, the amount of reflection of the deviation of the air-fuel ratio deviation index value AFF to the vapor concentration correction coefficient FGPG can be reduced, and erroneous learning of the vapor concentration correction coefficient FGPG can be prevented.

  In the present embodiment, the “learning method correcting means” according to the first, sixth and thirteenth inventions is realized by the ECU 40 executing the routine shown in FIG.

Embodiment 10 FIG.
Next, Embodiment 10 of the present invention will be described with reference to FIG. The system of the present embodiment is realized by causing the ECU 40 to execute a routine shown in FIG. 12 described later in place of the routine shown in FIG. 3 described above in the system of the first embodiment.

[Features of air-fuel ratio control according to the tenth embodiment]
Also in the air-fuel ratio control according to the present embodiment, regarding the learning of the fuel concentration, correction control of the learning method is further performed as described below based on the concentration learning routine shown in the flowchart of FIG. As in the correction control according to the ninth embodiment, the correction control according to the present embodiment increases the reference number KCS for the determination in step S6 of the concentration learning routine according to the low stability of the fuel adhesion state. Take the method.

  In the correction control according to the present embodiment, the cooling water temperature of the internal combustion engine 2 is acquired as a parameter related to the stability of the fuel adhesion state, as in the second embodiment. Then, the reference number KCS is increased according to the low coolant temperature. The adhesion state of the wall-attached fuel is particularly unstable when cold, and becomes stable as the internal combustion engine warms up. Therefore, if the reference number KCS is increased according to the low coolant temperature and the renewal period of the vapor concentration correction coefficient FGPG is extended, erroneous learning during cold weather can be effectively prevented. It is done.

[Specific Processing in Embodiment 10]
In the present embodiment, the update period of the vapor concentration correction coefficient FGPG is extended according to the flowchart of FIG. The flowchart of FIG. 12 shows a routine of update cycle extension control as correction control of the learning method. The routine shown in FIG. 12 is executed by the ECU 40, which is a control device for the internal combustion engine 2, together with the routine shown in FIG.

  In step S320 of the routine shown in FIG. 12, the reference number KCS is determined based on the coolant temperature THW. As shown in the figure, the reference number KCS is set to be large while the cooling water temperature THW is low, and is set to a smaller value as the cooling water temperature THW becomes higher. This is because as the cooling water temperature THW increases, that is, as the warm-up of the internal combustion engine 2 progresses, the wall surface temperature of the intake passage 6 also increases and the adhesion state of the wall-attached fuel becomes stable.

  The reference number KCS determined in the routine shown in FIG. 12 is taken into the routine shown in FIG. 2 and used for the determination in step S6.

  During cold weather, the adhesion state of the fuel adhering to the wall surface becomes unstable, and the air-fuel ratio tends to deviate from the target air-fuel ratio due to the fuel adhering to the wall surface and the fuel desorption from the wall surface. In such a situation, the deviation of the air-fuel ratio deviation index value AFF from the reference value increases with the deviation of the air-fuel ratio. The deviation of the air-fuel ratio deviation index value AFF is reflected in the vapor concentration correction coefficient FGPG. However, according to the routine shown in FIG. 12, the reference number KCS is increased and the update period of the vapor concentration correction coefficient FGPG is extended. Thus, the amount of reflection of the deviation of the air-fuel ratio deviation index value AFF to the vapor concentration correction coefficient FGPG can be reduced, and erroneous learning of the vapor concentration correction coefficient FGPG can be prevented.

  In the present embodiment, the “learning method correcting means” according to the first, sixth, and fourteenth inventions is realized by the ECU 40 executing the routine shown in FIG.

Embodiment 11 FIG.
Next, an eleventh embodiment of the present invention will be described with reference to FIG. The system of the present embodiment is realized by causing the ECU 40 to execute a routine shown in FIG. 13 described later in place of the routine shown in FIG. 3 described above in the system of the first embodiment.

[Features of air-fuel ratio control according to the eleventh embodiment]
Also in the air-fuel ratio control according to the present embodiment, regarding the learning of the fuel concentration, correction control of the learning method is further performed as described below based on the concentration learning routine shown in the flowchart of FIG. As in the correction control according to the ninth embodiment, the correction control according to the present embodiment increases the reference number KCS for the determination in step S6 of the concentration learning routine according to the low stability of the fuel adhesion state. Take the method.

  In the correction control according to the present embodiment, the elapsed time after returning from the fuel cut is acquired as a parameter related to the stability of the fuel adhesion state, as in the fourth embodiment. Then, the reference number KCS is increased in accordance with the short elapsed time after the return. Fuel adhesion to the wall becomes more noticeable as the elapsed time after returning from the fuel cut is shorter, so the reference count KCS is increased according to the shorter elapsed time and the vapor concentration correction factor FGPG update cycle is extended. If this is done, it is considered that erroneous learning at the time of return from the fuel cut can be effectively prevented.

[Specific Processing in Embodiment 11]
In the present embodiment, the update cycle of the vapor concentration correction coefficient FGPG is extended according to the flowchart of FIG. The flowchart of FIG. 13 shows a routine of update cycle extension control as correction control of the learning method. The routine shown in FIG. 13 is executed by the ECU 40, which is a control device for the internal combustion engine 2, together with the routine shown in FIG.

  In step S330 of the routine shown in FIG. 13, the reference number KCS is determined based on the elapsed time after returning from the fuel cut. As shown in the figure, the reference number KCS is set to a smaller value as the elapsed time after returning from the fuel cut becomes longer, and is set to a constant value when the elapsed time exceeds a predetermined time. This is because the fuel adheres to the wall surface as time elapses after returning from the fuel cut, and the wall surface fuel adheres to a stable state. Further, the temperature (cooling water temperature) of the internal combustion engine 2 is also considered in setting the reference number of times KCS. The higher the temperature of the internal combustion engine 2, the more stable the adhesion state of the wall-attached fuel. Therefore, even if the elapsed time after return is the same, the reference number KCS is set to a small value if the temperature of the internal combustion engine 2 is high.

  The reference number KCS determined in the routine shown in FIG. 13 is taken into the routine shown in FIG. 2 and used for the determination in step S6.

  After returning from the fuel cut, the fuel adheres to the wall surface so as to compensate for the desorption during the fuel cut, so that the air-fuel ratio tends to become lean. In such a situation, the deviation of the air-fuel ratio deviation index value AFF in the positive direction increases as the air-fuel ratio becomes leaner. The deviation of the air-fuel ratio deviation index value AFF is reflected in the vapor concentration correction coefficient FGPG. However, according to the routine shown in FIG. 13, the reference number KCS is increased and the update period of the vapor concentration correction coefficient FGPG is extended. Thus, the amount of reflection of the deviation of the air-fuel ratio deviation index value AFF to the vapor concentration correction coefficient FGPG can be reduced, and erroneous learning of the vapor concentration correction coefficient FGPG can be prevented.

  In the present embodiment, the “learning method correcting means” according to the first, sixth, fourteenth and fifteenth inventions is realized by the ECU 40 executing the routine shown in FIG.

Embodiment 12 FIG.
Next, Embodiment 12 of the present invention will be described with reference to FIG. The system of the present embodiment is realized by causing the ECU 40 to execute a routine shown in FIG. 14 described later in place of the routine shown in FIG. 3 described above in the system of the first embodiment.

[Features of air-fuel ratio control according to the twelfth embodiment]
Also in the air-fuel ratio control according to the present embodiment, regarding the learning of the fuel concentration, correction control of the learning method is further performed as described below based on the concentration learning routine shown in the flowchart of FIG. As in the correction control according to the ninth embodiment, the correction control according to the present embodiment increases the reference number KCS for the determination in step S6 of the concentration learning routine according to the low stability of the fuel adhesion state. Take the method.

  In the correction control according to the present embodiment, as in the fifth embodiment, the residual degree of the influence of the fuel increase is acquired as a parameter related to the stability of the fuel adhesion state. Then, the reference number KCS is increased according to the degree of the residual degree. The fuel desorption from the wall surface becomes more prominent as the amount of fuel adhering to the wall surface increases, that is, as the effect of fuel increase remains, so the reference number KCS is increased according to the residual degree of the effect of fuel increase. If the renewal period of the vapor concentration correction coefficient FGPG is extended, it is considered that erroneous learning at the time of return from the fuel increase can be effectively prevented.

[Specific Processing in Embodiment 12]
In the present embodiment, the update cycle of the vapor concentration correction coefficient FGPG is extended according to the flowchart of FIG. The flowchart of FIG. 14 shows a routine of update cycle extension control as correction control of the learning method. The routine shown in FIG. 14 is executed by the ECU 40, which is a control device for the internal combustion engine 2, together with the routine shown in FIG.

  In the first step S340 of the routine shown in FIG. 14, first, the increase coefficient KRICHX is smoothed in the time direction by the above equation (17). Further, the deviation ΔRICH between the increase coefficient smoothing value RICHSM and the current increase coefficient KRICHX is calculated by the above equation (18).

  In the next step S342, the reference number KCS is determined based on the deviation ΔRICH calculated in step S340. As shown in the drawing, the reference number KCS is set to a fixed number (first predetermined number) until the deviation ΔRICH exceeds a predetermined value. However, when the deviation ΔRICH exceeds a predetermined value, the reference count KCS is set to a larger value as the deviation ΔRICH is larger. This is because as the deviation ΔRICH is larger, the amount of fuel attached to the wall surface becomes excessive with respect to the stable amount, and fuel desorption from the wall surface proceeds. When the deviation ΔRICH increases to some extent, the reference number KCS is held at a second predetermined number of times that is greater than the first predetermined number of times.

  The reference number KCS determined in the routine shown in FIG. 14 is taken into the routine shown in FIG. 2 and used for the determination in step S6.

  After returning from the fuel increase, the fuel adsorbs excessively during the fuel increase, so that the fuel desorbs from the wall surface, so that the air-fuel ratio tends to become rich. In such a situation, the deviation of the air-fuel ratio deviation index value AFF in the negative direction increases with the enrichment of the air-fuel ratio. The deviation of the air-fuel ratio deviation index value AFF is reflected in the vapor concentration correction coefficient FGPG. However, according to the routine shown in FIG. 14, the reference number KCS is increased and the vapor concentration correction coefficient FGPG is increased when the influence of the fuel increase is large. By extending the update period, it is possible to reduce the reflection amount of the deviation of the air-fuel ratio deviation index value AFF to the vapor concentration correction coefficient FGPG, and to prevent erroneous learning of the vapor concentration correction coefficient FGPG.

  In the present embodiment, the “learning method correcting means” according to the first, sixth and sixteenth inventions is realized by the ECU 40 executing the routine shown in FIG.

Embodiment 13 FIG.
Next, a thirteenth embodiment of the present invention will be described with reference to FIG. The system of the present embodiment is realized by causing the ECU 40 to execute a routine shown in FIG. 15 described later in place of the routine shown in FIG. 3 described above in the system of the first embodiment.

[Features of air-fuel ratio control according to Embodiment 13]
Also in the air-fuel ratio control according to the present embodiment, regarding the learning of the fuel concentration, correction control of the learning method is further performed as described below based on the concentration learning routine shown in the flowchart of FIG. Unlike the correction control according to the first embodiment, the correction control according to the present embodiment adopts a method of interrupting the update of the vapor concentration correction coefficient FGPG when the stability of the fuel adhesion state is low. The updating of the vapor concentration correction coefficient FGPG can be interrupted by stopping or forcibly resetting the learning counter CFGPG in step S4. By interrupting the update, it is possible to prevent the deviation of the air-fuel ratio due to the change in the amount of fuel attached to the wall surface from being taken into the vapor concentration correction coefficient FGPG.

  In the correction control according to the present embodiment, the deviation ΔWP between the basic adhesion rate WPBS and the smooth value WP of the basic adhesion rate WPBS is acquired as a parameter related to the stability of the fuel adhesion state, as in the first embodiment. Then, based on the value of the deviation ΔWP, it is determined whether to interrupt the update of the vapor concentration correction coefficient FGPG. By performing the correction control with the deviation ΔWP as the parameter, it is possible to accurately grasp the fuel adhesion state on the wall surface or the fuel desorption state from the wall surface to determine whether the update is interrupted. The influence which the change of quantity has on a learning value can be suppressed effectively.

[Specific Processing in Embodiment 13]
In the present embodiment, updating of the vapor concentration correction coefficient FGPG is interrupted according to the flowchart of FIG. The flowchart of FIG. 15 shows a routine of update interruption control as correction control of the learning method. The routine shown in FIG. 15 is executed by the ECU 40 that is a control device for the internal combustion engine 2 together with the routine shown in FIG.

  In the first step S400 of the routine shown in FIG. 15, the basic adhesion rate WPBS is calculated by the above equation (9). The basic adhesion rate WPBS has a meaning as a parameter indicating the stable adhesion state of the wall-attached fuel.

  In the next step S402, it is determined whether or not a fuel cut is being executed. If it is determined that the fuel cut is being executed, the basic adhesion rate WPBS is set to zero in step S404.

  In the next step S406, a smoothing coefficient n used for smoothing the basic adhesion rate WPBS is determined based on the cooling water temperature THW. The smoothing coefficient n is a value of 1 or more, and is set to a smaller value as the cooling water temperature THW is higher, as shown in the figure.

  In the next step S408, using the smoothing coefficient n determined in step S406, the basic adhesion rate WPBS is smoothed in the time direction according to the above equation (10). The basic adhesion rate smooth value WP calculated here has a meaning as a parameter indicating the current adhesion state of the wall-attached fuel.

  In the next step S410, as shown in the above equation (11), a deviation ΔWP between the basic adhesion rate smooth value WP calculated in step S408 and the basic adhesion rate WPBS calculated in step S400 or step S404 is calculated. Is done. This deviation ΔWP has a meaning as a parameter related to the stability of the fuel adhesion state.

  The processes in steps S412 and S414 are processes for determining whether or not the deviation ΔWP calculated in step S410 exceeds the allowable range. In step S412, it is determined whether or not the deviation ΔWP is greater than or equal to the upper limit value KWP of the allowable range. If the deviation ΔWP is less than the upper limit value KWP of the allowable range, the process proceeds to step S414, and it is determined whether or not the deviation ΔWP is less than or equal to the lower limit value KWP of the allowable range. As a result of these determinations, when the deviation ΔWP is within the allowable range, a flag permitting the update of the vapor concentration correction coefficient FGPG is set by the processing of step S418. On the other hand, when the deviation ΔWP exceeds the allowable range, a flag that prohibits the update of the vapor concentration correction coefficient FGPG is set by the process of step S416.

  The determination result of the update interruption made by the routine shown in FIG. 15 is taken into the routine shown in FIG. 2 and reflected in the process of step S4. Specifically, when the update permission flag is set by the process of step S418, the learning counter CFGPG is incremented in step S4. On the other hand, when the update non-permission flag is set by the process of step S416, the increment of the learning counter CFGPG is stopped or forcibly reset to zero in step S4. As a result, the update of the vapor concentration correction coefficient FGPG is interrupted.

  In the situation where the fuel adheres to the wall surface as in the purge cut, the air-fuel ratio deviation index value AFF is shifted in the positive direction as the air-fuel ratio becomes leaner. On the other hand, in the situation where fuel desorption proceeds from the wall surface at the start of the purge, the air-fuel ratio deviation index value AFF is shifted in the negative direction as the air-fuel ratio is enriched. The deviation of the air-fuel ratio deviation index value AFF is reflected in the vapor concentration correction coefficient FGPG. However, according to the routine shown in FIG. 15, the update of the vapor concentration correction coefficient FGPG is interrupted in the situation where the deviation of the air-fuel ratio deviation index value AFF is large. Therefore, the deviation of the air-fuel ratio deviation index value AFF is not reflected in the vapor concentration correction coefficient FGPG, and erroneous learning of the vapor concentration correction coefficient FGPG can be prevented.

  In the present embodiment, the “learning method correcting means” according to the first, seventh and thirteenth inventions is realized by the ECU 40 executing the routine shown in FIG.

Embodiment 14 FIG.
Next, a fourteenth embodiment of the present invention will be described with reference to FIG. The system of the present embodiment is realized by causing the ECU 40 to change a part of the routine shown in FIG. 15 described above to a routine shown in FIG. .

[Features of air-fuel ratio control according to Embodiment 14]
According to the routine shown in FIG. 15, if the deviation ΔWP is within the allowable range again when the update of the vapor concentration correction coefficient FGPG is interrupted, the update of the vapor concentration correction coefficient FGPG is resumed at that time. The However, since there is a gas transport delay from the intake passage 6 to the exhaust gas sensor 34, there is some time lag before the change in the amount of fuel attached to the wall surface appears in the output of the exhaust gas sensor 34. In the present embodiment, in consideration of this time lag, the vapor concentration correction coefficient FGPG is updated again after a certain amount of time has elapsed after the deviation ΔWP has entered the allowable range again. According to this, the influence which the change in the amount of fuel adhering to the wall surface has on the learned value can be suppressed more reliably.

[Specific Processing in Embodiment 14]
In the present embodiment, updating of the vapor concentration correction coefficient FGPG is interrupted according to the flowchart of FIG. The flowchart in FIG. 16 is a part of the update interruption control routine, and step S412 and subsequent steps of the routine shown in FIG. 15 are modified and executed as in the routine shown in FIG.

  First, the processing in steps S420 and S422 of the routine shown in FIG. 16 is processing for determining whether or not the deviation ΔWP calculated in step S410 of the routine shown in FIG. 15 exceeds the allowable range. In step S420, it is determined whether or not the deviation ΔWP is equal to or greater than the upper limit value + KWP of the allowable range. If the deviation ΔWP is less than the upper limit value + KWP of the allowable range, the process proceeds to step S422, and it is determined whether or not the deviation ΔWP is less than or equal to the lower limit value −KWP of the allowable range.

  As a result of the determinations in steps S420 and S422, if the deviation ΔWP is within the allowable range, the value of the elapsed time counter CWP is increased by a certain value by the processing in step S426. This elapsed time counter CWP corresponds to the elapsed time after the deviation ΔWP is within the allowable range. If the deviation ΔWP exceeds the allowable range, the elapsed time counter CWP is reset to zero by the process of step S424.

  In the next step S428, the reference number KTWPWP of the elapsed time counter CWP is read. In the present embodiment, the reference number KTWPWP is fixed to a constant value (adapted value). In step S430, it is determined whether or not the elapsed time counter CWP has reached the reference number KTWPWP. Until the elapsed time counter CWP reaches the reference number KTWPWP, a flag that prohibits the update of the vapor concentration correction coefficient FGPG is set by the processing of step S434. On the other hand, when the elapsed time counter CWP reaches the reference number KTWPWP, a flag for permitting the update of the vapor concentration correction coefficient FGPG is set by the process of step S432.

  The determination result of the update interruption made by the routine shown in FIG. 16 is taken into the routine shown in FIG. 2 and reflected in the process of step S4. While the update disapproval flag is set by the process of step S434, the increment of the learning counter CFGPG is stopped or forcibly reset to zero in step S4. On the other hand, when the update permission flag is set by the process of step S432, the increment of the learning counter CFGPG is resumed in step S4, and the update of the vapor concentration correction coefficient FGPG is resumed.

  In the present embodiment, the ECU 40 changes a part of the routine shown in FIG. 15 and executes the routine shown in FIG. 16, so that the “learning method” according to the first, seventh, eighth, and thirteenth inventions is applied. "Correction means" is realized.

Embodiment 15 FIG.
Next, Embodiment 15 of the present invention will be described with reference to FIG. The system of the present embodiment is realized by causing the ECU 40 to change and execute a part of the routine shown in FIG. 16 as shown in FIG. 17 to be described later in the system of the fourteenth embodiment. .

[Features of air-fuel ratio control according to the fifteenth embodiment]
In the routine shown in FIG. 16, the reference number KTWPWP serving as a reference for restarting the update of the vapor concentration correction coefficient FGPG is fixed to a constant value. On the other hand, in the present embodiment, the coolant temperature of the internal combustion engine 2 is acquired, and the reference number KTWPWP is increased according to the low coolant temperature. The adhesion state of the wall-attached fuel is particularly unstable when cold, and becomes stable as the internal combustion engine warms up. Therefore, if the reference number KTWPWP is increased according to the low coolant temperature and the time until the renewal of the vapor concentration correction coefficient FGPG is increased, the change in the amount of fuel adhering to the wall during the cold period will be the learned value. It is thought that the influence which it gives can be suppressed.

[Specific Processing in Embodiment 15]
In the present embodiment, the reference number KTWPWP is set according to the flowchart of FIG. The flowchart in FIG. 17 is a part of the update interruption control routine, and step S428 of the routine shown in FIG. 16 is modified and executed as in the routine shown in FIG.

  In step S440 of the routine shown in FIG. 17, the reference number KTWPWP is determined based on the cooling water temperature THW. As shown in the figure, the reference number KTWPWP is large when the cooling water temperature THW is low, and is set to a small value as the cooling water temperature THW increases. This is because as the cooling water temperature THW increases, that is, as the warm-up of the internal combustion engine 2 progresses, the wall surface temperature of the intake passage 6 also increases and the adhesion state of the wall-attached fuel becomes stable. Further, the rotational speed NE of the internal combustion engine 2 is also taken into consideration in setting the reference number KTWPWP. The larger the rotational speed NE, the smaller the gas transport delay. Therefore, even if the cooling water temperature THW is the same, if the rotational speed NE is high, the reference number KTWPWP is set to a small value.

  The reference number KTWPWP determined by the routine shown in FIG. 17 is taken into the routine shown in FIG. 16 and used for the determination in step S430.

Embodiment 16 FIG.
Next, Embodiment 16 of the present invention will be described with reference to FIG. The system according to the present embodiment is realized by causing the ECU 40 to change a part of the routine shown in FIG. 16 described above to a routine shown in FIG. .

[Features of air-fuel ratio control according to the sixteenth embodiment]
In the routine shown in FIG. 16, the reference number KTWPWP serving as a reference for restarting the update of the vapor concentration correction coefficient FGPG is fixed to a constant value. In contrast, in the present embodiment, the elapsed time after returning from the fuel cut is acquired, and the reference number KTWPWP is increased according to the shortness of the elapsed time after returning. The adhesion of fuel to the wall surface becomes more prominent as the elapsed time after returning from the fuel cut is shorter. Therefore, if the reference number KTWPWP is increased in accordance with the shortness of the elapsed time after recovery and the time until restart of the vapor concentration correction coefficient FGPG is increased, the amount of fuel adhering to the wall at the time of recovery from the fuel cut It is considered that the influence of the change in the learning value can be suppressed.

[Specific Processing in Sixteenth Embodiment]
In the present embodiment, the reference number KTWPWP is set according to the flowchart of FIG. The flowchart in FIG. 18 is a part of the update interruption control routine, and step S428 of the routine shown in FIG. 16 is changed to the routine shown in FIG.

  In step S450 of the routine shown in FIG. 18, the reference number KTWPWP is determined based on the elapsed time after returning from the fuel cut. As shown in the figure, the reference number KTWPWP is set to a smaller value as the elapsed time after returning from the fuel cut becomes longer. This is because the fuel adheres to the wall surface as time elapses after returning from the fuel cut, and the wall surface fuel adheres to a stable state. Further, the temperature (cooling water temperature) of the internal combustion engine 2 is also considered in setting the reference number KTWPWP. As the temperature of the internal combustion engine 2 is higher, the adhesion state of the wall-attached fuel becomes more stable. Therefore, even if the elapsed time after the return is the same, the reference number KTWPWP is set to a smaller value if the temperature of the internal combustion engine 2 is higher.

  The reference number KTWPWP determined by the routine shown in FIG. 18 is taken into the routine shown in FIG. 16 and used for the determination in step S430.

Embodiment 17. FIG.
Next, Embodiment 17 of the present invention will be described with reference to FIG. The system of the present embodiment is realized by causing the ECU 40 to change a part of the routine shown in FIG. 16 described above to a routine shown in FIG. .

[Features of air-fuel ratio control according to the seventeenth embodiment]
In the routine shown in FIG. 16, the reference number KTWPWP serving as a reference for restarting the update of the vapor concentration correction coefficient FGPG is fixed to a constant value. In contrast, in the present embodiment, the residual degree of the influence of the fuel increase is acquired, and the reference number KTWPWP is increased according to the residual degree. The fuel desorption from the wall surface becomes more prominent as the amount of fuel adhering to the wall surface is excessive, that is, as the influence of the fuel increase remains. Therefore, if the reference number KTWPWP is increased according to the residual level of the effect of fuel increase and the time until restart of the vapor concentration correction coefficient FGPG is increased, the amount of fuel adhering to the wall at the time of return from fuel increase will be reduced. It is thought that the influence that the change has on the learning value can be suppressed.

[Specific Processing in Embodiment 17]
In the present embodiment, the reference number KTWPWP is set according to the flowchart of FIG. The flowchart in FIG. 19 is a part of the update interruption control routine, and step S428 of the routine shown in FIG. 16 is modified and executed as in the routine shown in FIG.

  In the first step S460 of the routine shown in FIG. 19, first, the increase coefficient KRICHX is smoothed in the time direction by the equation (17). Further, the deviation ΔRICH between the increase coefficient smoothing value RICHSM and the current increase coefficient KRICHX is calculated by the above equation (18).

  In the next step S462, the reference number KTWPWP is determined based on the deviation ΔRICH calculated in step S460. As shown in the figure, the reference number KTWPWP is set to a fixed number (first predetermined number) until the deviation ΔRICH exceeds a predetermined value. However, when the deviation ΔRICH exceeds a predetermined value, the reference count KTWPWP is set to a larger value as the deviation ΔRICH is larger. This is because as the deviation ΔRICH is larger, the amount of fuel attached to the wall surface becomes excessive with respect to the stable amount, and fuel desorption from the wall surface proceeds. When the deviation ΔRICH increases to some extent, the reference number KTWPWP is held at a second predetermined number of times that is greater than the first predetermined number of times.

  The reference number KTWPWP determined by the routine shown in FIG. 19 is taken into the routine shown in FIG. 16 and used for the determination in step S430.

Embodiment 18 FIG.
Next, an eighteenth embodiment of the present invention will be described with reference to FIG. The system of the present embodiment is realized by causing the ECU 40 to execute a routine shown in FIG. 20 described later in place of the routine shown in FIG. 3 described above in the system of the first embodiment.

[Features of air-fuel ratio control according to Embodiment 18]
Also in the air-fuel ratio control according to the present embodiment, regarding the learning of the fuel concentration, correction control of the learning method is further performed as described below based on the concentration learning routine shown in the flowchart of FIG. Unlike the correction control according to the first embodiment, the correction control according to the present embodiment calculates a correction value for offsetting the influence of the change in the wall-attached fuel amount on the feedback correction coefficient FAF. Is reflected in the air-fuel ratio deviation index value AFF used in the concentration learning routine. If the influence of the change in the amount of fuel adhering to the wall surface on the feedback correction coefficient FAF is excluded in advance from the air-fuel ratio deviation index value AFF, the fuel adhering to the wall surface is added to the learning update amount DLFGPG calculated based on the air-fuel ratio deviation index value AFF. The effects of changes in quantity are not included. Therefore, according to the method employed in the present embodiment, the vapor concentration correction coefficient FGPG can be learned with high accuracy even when the fuel adhesion state is unstable.

  In the correction control according to the present embodiment, the deviation ΔWP between the basic adhesion rate WPBS and the smooth value WP of the basic adhesion rate WPBS is acquired as a parameter related to the stability of the fuel adhesion state, as in the first embodiment. Then, this deviation ΔWP is used as a correction value for offsetting the influence of the change in the amount of fuel attached to the wall surface on the feedback correction coefficient FAF. The deviation ΔWP is a parameter indicating the degree of deviation of the air-fuel ratio to the rich side due to the desorption of fuel from the wall when the value is positive, and when the value is negative, the deviation ΔWP is attached to the wall. This is a parameter indicating the degree of deviation of the air-fuel ratio to the lean side. Therefore, by using the deviation ΔWP as a correction value and adding it to the air-fuel ratio deviation index value AFF, the influence of the change in the amount of fuel adhering to the wall on the air-fuel ratio can be reliably excluded from the air-fuel ratio deviation index value AFF. .

[Specific Processing in Eighteenth Embodiment]
In the present embodiment, the air-fuel ratio deviation index value AFF is corrected in order to eliminate the influence of the change in the amount of fuel attached to the wall according to the flowchart of FIG. The flowchart of FIG. 20 shows a routine of influence deletion control as correction control of the learning method. The routine shown in FIG. 20 is executed by the ECU 40, which is a control device for the internal combustion engine 2, together with the routine shown in FIG.

  In the first step S500 of the routine shown in FIG. 20, the basic adhesion rate WPBS is calculated by the above equation (9). The basic adhesion rate WPBS has a meaning as a parameter indicating the stable adhesion state of the wall-attached fuel.

  In the next step S502, it is determined whether or not a fuel cut is being executed. If it is determined that the fuel cut is being executed, the basic adhesion rate WPBS is set to zero in step S504.

  In the next step S506, a smoothing coefficient n used for smoothing the basic adhesion rate WPBS is determined based on the cooling water temperature THW. The smoothing coefficient n is a value of 1 or more, and is set to a smaller value as the cooling water temperature THW is higher, as shown in the figure.

  In the next step S508, using the smoothing coefficient n determined in step S506, the basic adhesion rate WPBS is smoothed in the time direction according to the above equation (10). The basic adhesion rate smooth value WP calculated here has a meaning as a parameter indicating the current adhesion state of the wall-attached fuel.

  In the next step S510, as shown in the above equation (11), a deviation ΔWP between the basic adhesion rate smooth value WP calculated in step S508 and the basic adhesion rate WPBS calculated in step S500 or step S504 is calculated. Is done. This deviation ΔWP has a meaning as a parameter indicating the direction and degree of deviation of the air-fuel ratio in a situation where the adhesion state of the fuel adhered to the wall surface becomes unstable.

In step S512, the air-fuel ratio deviation index value AFF is calculated by the following equation (23) using the deviation ΔWP obtained in step S510. The air-fuel ratio deviation index value AFF calculated by the equation (23) is equivalent to the air-fuel ratio deviation index value AFF calculated by the above equation (4) plus the deviation ΔWP and the increase coefficient KRICH.
AFF = FAF + (ABYF / ABYF0) + ΔWP + KRICH (23)

  The air-fuel ratio deviation index value AFF calculated in the routine shown in FIG. 20 is taken into the routine shown in FIG. 2 and used for the calculation in step S2 and the determinations in steps S10 and S12.

  The feedback correction coefficient FAF reflects the deviation of the air-fuel ratio from the target air-fuel ratio, and in addition to the deviation of the air-fuel ratio due to the effect of the purge gas, the deviation of the air-fuel ratio caused by the change in the amount of fuel attached to the wall In addition, a deviation of the air-fuel ratio due to the influence of fuel increase is also included. According to the routine shown in FIG. 20, by adding the deviation ΔWP and the increase coefficient KRICH to the feedback correction coefficient FAF, the influence of factors other than the purge gas on the feedback correction coefficient FAF can be offset. Therefore, the air-fuel ratio deviation index value AFF calculated by the equation (23) reflects only the air-fuel ratio deviation due to the influence of the purge gas, and the learning update amount DLFGPG is calculated using this air-fuel ratio deviation index value AFF. By doing so, it is possible to improve the learning accuracy of the vapor concentration correction coefficient FGPG in a situation where the adhesion state of the wall surface fuel becomes unstable.

  In the present embodiment, the “learning method correcting means” according to the first, ninth and thirteenth inventions is realized by the ECU 40 executing the routine shown in FIG.

Embodiment 19. FIG.
Next, Embodiment 19 of the present invention will be described with reference to FIG. The system of the present embodiment is realized by causing the ECU 40 to change and execute a part of the routine shown in FIG. 20 as shown in the routine shown in FIG. 21 described later in the system of the twentieth embodiment. .

[Features of air-fuel ratio control according to Embodiment 19]
In the routine shown in FIG. 2, the concentration learning is performed only when the air-fuel ratio deviation index value AFF exceeds a predetermined dead band. Providing a dead band is effective in eliminating the influence of noise components and maintaining the stability of air-fuel ratio control. However, there is a problem that even if the air-fuel ratio deviation index value AFF is deviated due to the purge gas, the concentration learning does not proceed when the air-fuel ratio deviation index value AFF is within the dead band. Therefore, in the present embodiment, when the learning update amount DLFGPG is calculated from the air-fuel ratio deviation index value AFF, the learning update amount DLFGPG is calculated using the air-fuel ratio deviation index value AFF expressed by the following equation (24). I have to.
AFF = FAF + (ABYF / ABYF0) + ΔWP + KRICH + ΔDL (24)

  The air-fuel ratio deviation index value AFF calculated by the equation (24) corresponds to the air-fuel ratio deviation index value AFF calculated by the equation (23) further added with the padded value ΔDL. The inflated value ΔDL corresponds to the upper limit value KAFFH or the lower limit value KAFFL of the dead band, and is set according to the direction of deviation of the air-fuel ratio deviation index value AFF. By adding the inflated value ΔDL, it is possible to increase the learning update amount DLFGPG when the concentration learning is performed when the air-fuel ratio deviation index value AFF exceeds the dead band, and the vapor concentration correction coefficient FGPG The learning speed can be increased.

[Specific Processing in Embodiment 19]
In the present embodiment, the air-fuel ratio deviation index value AFF is corrected in order to eliminate the influence of the change in the wall surface attached fuel amount according to the flowchart of FIG. The flowchart of FIG. 21 is a part of the influence deletion control routine as the correction control of the learning method, and is executed by changing S512 of the routine shown in FIG. 20 to the routine shown in FIG.

First, in step S520 of the routine shown in FIG. 21, a correction direction index value AFFO that is a parameter indicating the correction direction of the air-fuel ratio is calculated by the following equation (25). The correction direction index value AFFO corresponds to a value obtained by adding the increase coefficient KRICH to the air-fuel ratio deviation index value AFF calculated by the above equation (4). By adding the increase coefficient KRICH, the effect of fuel increase is eliminated. Therefore, the correction direction index value AFFO reflects only the air-fuel ratio shift due to the influence of the purge gas and the air-fuel ratio shift caused by the change in the amount of fuel attached to the wall surface.
AFFO = FAF + (ABYF / ABYF0) + KRICH (25)

In step S522, the correction direction index value AFFO is smoothed in the time direction by the following equation (26). In the equation (26), AFFOSM is a smooth value of the correction direction index value AFFO, AFFOSM on the left side is a smoothed value after update, and AFFOSM on the right side is a smooth value before update. NA is a smoothing coefficient.
AFFOSM = AFFOSM + (AFFO-AFFOSM) / NA (26)

  In the next step S524, it is determined whether or not the density update count CFGCNG counted in step S32 of the routine shown in FIG. 2 is equal to or greater than a predetermined reference count KCNG0. If the density update count CFGCNG exceeds the reference count KCNG0, it is determined that the density learning is sufficiently advanced. On the other hand, if the concentration update count CFGCNG is less than the reference count KCNG0, it is determined that it is necessary to advance the concentration learning by increasing the learning speed of the vapor concentration correction coefficient FGPG.

  If the result of determination in step S524 is that the density update count CFGCNG is greater than or equal to the reference count KCNG0, the process of step S530 is performed. In step S530, the above-described padding value ΔDL is set to a predetermined value KΔDL0. If step S530 is selected, the concentration learning has already progressed sufficiently, so that it is not necessary to increase the air-fuel ratio deviation index value AFF in order to increase the learning speed. Therefore, the predetermined value KΔDL0 is preferably set to zero.

  If the result of determination in step S526 is that the density update count CFGCNG is less than the reference count KCNG0, determinations in steps S526 and S528 are performed. These determinations are processes for detecting the correction direction from the correction direction index smooth value AFFO calculated in step S522. Here, the range of −KDFPG to + KDFPG is set as a dead band in which the padding correction by the padding value ΔDL is not performed. The dead band is set in consideration of errors due to noise and the like. In step S526, it is determined whether or not the correction direction index smoothing value AFFOSM is equal to or greater than the upper limit of the dead band + KDFPG. If the correction direction index smooth value AFFOSM is less than the upper limit value + KDFPG, the process proceeds to step S528, and it is determined whether or not the correction direction index smooth value AFFOSM is less than or equal to the lower limit value -KDFPGKWP.

  As a result of the determination in steps S526 and S528, if the correction direction index smooth value AFFOSM is equal to or less than the lower limit value −KDFPG of the dead band, the process of step S532 is performed. In step S532, the above-described padding value ΔDL is set to a predetermined value KΔDL1. When step S532 is selected, the air-fuel ratio is shifted to the rich side, so the value of the air-fuel ratio shift index value AFF may be increased in the rich direction, that is, the minus direction. By doing so, the learning update amount DLFGPG can be increased in the minus direction, that is, in the high concentration direction, and accordingly, the learning speed of the vapor concentration correction coefficient FGPG in the high concentration direction can be increased. Therefore, the predetermined value KΔDL1 is preferably set to a negative value, particularly a lower limit value KAFFL of the dead band set in step S12 of the routine shown in FIG.

  On the other hand, if the result of determination in steps S526 and S528 is that the correction direction index smooth value AFFOSM is equal to or greater than the upper limit value of the dead band + KDFPG, the process of step S534 is performed. In step S534, the above-described padding value ΔDL is set to a predetermined value KΔDL2. When step S534 is selected, the air-fuel ratio is deviating toward the lean side, so the value of the air-fuel ratio deviation index value AFF may be increased in the lean direction, that is, in the plus direction. By doing so, the learning update amount DLFGPG can be increased in the positive direction, that is, in the low concentration direction, and the learning speed of the vapor concentration correction coefficient FGPG in the low concentration direction can be increased accordingly. Therefore, it is preferable to set the predetermined value KΔDL2 to a positive value, in particular, to the upper limit value KAFFH of the dead band set in step S10 of the routine shown in FIG.

  If the correction direction index smooth value AFFOSM is within the dead band as a result of the determinations in steps S526 and S528, the direction of correction for increasing the air-fuel ratio deviation index value AFF is not determined. Therefore, the value of the above-described padding value ΔDL is set to the predetermined value KΔDL0, that is, zero by the process of step S530.

  In step S536, the air-fuel ratio deviation index value AFF is calculated by the above-described equation (24) using the inflated value ΔDL set in any of steps S530, S532, and S534. The air-fuel ratio deviation index value AFF calculated in step S536 is taken into the routine shown in FIG. 2 and used for the calculation in step S2. In the determinations in steps S10 and S12, not the air-fuel ratio deviation index value AFF calculated by the equation (24) but the air-fuel ratio deviation index value AFF calculated by the equation (23), that is, the padding correction is performed. The previous air-fuel ratio deviation index value AFF is used.

  According to the routine shown in FIG. 21, when the air-fuel ratio is shifted to the lean side, the padding value ΔDL set to be equal to or higher than the upper limit value KAFFH of the dead band is added to the feedback correction coefficient FAF, so that FIG. The learning update amount DLFGPG calculated by the routine shown in FIG. Conversely, when the air-fuel ratio is shifted to the rich side, the inflated value ΔDL set below the lower limit value KAFFL of the dead band is added to the feedback correction coefficient FAF, so that the learning update amount DLFGPG is increased in the high concentration direction. Can be inflated. According to this, it is possible to increase the learning speed while providing the dead band, and it is possible to increase the learning accuracy of the vapor concentration correction coefficient FGPG at an early stage.

  Further, according to the routine shown in FIG. 21, as in the routine shown in FIG. 20, the deviation ΔWP and the increase coefficient KRICH are also added to the feedback correction coefficient FAF, so that the factors other than the purge gas affect the feedback correction coefficient FAF. Is offset. As a result, it is possible to obtain the learning update amount DLFGPG that reflects only the deviation of the air-fuel ratio due to the effect of the purge gas, and improves the learning accuracy of the vapor concentration correction coefficient FGPG in the situation where the adhesion state of the fuel adhering to the wall surface becomes unstable Can be made.

  In the present embodiment, the ECU 40 changes a part of the routine shown in FIG. 20 and executes the routine shown in FIG. 21, whereby the first, ninth, tenth and thirteenth inventions are applied. "Learning method correction means" is realized.

Embodiment 20. FIG.
Next, a twentieth embodiment of the present invention will be described with reference to FIGS. The system of the present embodiment is realized by causing the ECU 40 to change and execute a part of the routine shown in FIG. 20 described above to a routine shown in FIG. 22 described later in the system of the twentieth embodiment. .

[Features of air-fuel ratio control according to Embodiment 20]
In the routine shown in FIG. 21, the learning update amount DLFGPG is increased in the direction corresponding to the concentration of the fuel concentration by adding the water increase value ΔDL corresponding to the direction of the air / fuel ratio shift to the air / fuel ratio shift index value AFF. This increases the learning speed of the vapor concentration correction coefficient FGPG. However, the fuel concentration of the purge gas depends on the amount of evaporated fuel adsorbed on the canister 18, and the amount of evaporated fuel adsorbed on the canister 18 depends on the amount of fuel evaporated in the fuel tank 20. For this reason, when the evaporation state of the fuel in the fuel tank 20 is unstable, the fuel concentration of the purge gas is likely to be unstable. Specifically, this is when the temperature in the fuel tank 20 is high. In such a situation, if the padding correction of the air-fuel ratio deviation index value AFF is performed, there is a possibility that the accuracy of the concentration degree learning may be lowered instead. Therefore, in the present embodiment, when the temperature in the fuel tank 20 is high, the correction for increasing the air-fuel ratio deviation index value AFF is prohibited.

  Further, the amount of evaporated fuel adsorbed on the canister 18 gradually decreases with the purge, and the fuel concentration of the purge gas also decreases accordingly. Therefore, when the integrated amount of purge increases to a certain extent, it is preferable to increase the air-fuel ratio deviation index value AFF in the lean direction, that is, in the plus direction. By doing so, it is possible to increase the learning update amount DLFGPG in the low concentration direction and increase the learning speed of the vapor concentration correction coefficient FGPG in the low concentration direction. Therefore, in the present embodiment, when the concentration update count CFGCNG counted in step S32 of the routine shown in FIG. 2 reaches a predetermined count, the value of the air-fuel ratio deviation index value AFF used for calculating the learning update amount DLFGPG is set. It is supposed to increase in the positive direction.

[Specific Processing in Embodiment 20]
In the present embodiment, the air-fuel ratio deviation index value AFF is corrected in order to eliminate the influence of the change in the wall surface attached fuel amount according to the flowchart of FIG. The flowchart of FIG. 22 is a part of the influence deletion control routine as the correction control of the learning method, and is executed by changing S512 of the routine shown in FIG. 20 to the routine shown in FIG.

  First, in step S540 of the routine shown in FIG. 22, the correction direction index value AFFO, which is a parameter indicating the correction direction of the air-fuel ratio, is calculated by the above equation (25). In the subsequent step S542, the correction direction index value AFFO is smoothed in the time direction by the above equation (26).

  In step S544, it is determined whether the temperature THA in the fuel tank 20 is equal to or higher than a predetermined reference temperature KTH. In a situation where the tank temperature THA is equal to or higher than the reference temperature KTH, the fuel evaporation state in the fuel tank 20 is unstable, and the fuel concentration of the purge gas from the canister 18 is likely to be unstable. . Therefore, when the tank temperature THA is equal to or higher than the reference temperature KTH, the process of step S554 is performed. In step S554, the inflated value ΔDL in the above equation (24) is set to a predetermined value KΔDL0. The predetermined value KΔDL0 is set to zero as described above.

  If the tank internal temperature THA is lower than the reference temperature KTH, the determination in step S546 is then performed. In step S546, it is determined whether or not the density update count CFGCNG counted in step S32 of the routine shown in FIG. 2 is equal to or greater than a predetermined reference count KCNG2. If the density update count CFGCNG exceeds the reference count KCNG2, it can be determined that the density learning is sufficiently advanced. Therefore, when the concentration update count CFGCNG exceeds the reference count KCNG2, the process of step S554 is performed, and the value of the padding value ΔDL in the above equation (24) is set to a predetermined value KΔDL0.

  If the density update count CFGCNG is less than the reference count KCNG2, then the determination in step S548 is performed. In step S548, it is determined whether the density update count CFGCNG is equal to or less than a predetermined reference count KCNG1. The reference number KCNG1 is smaller than the reference number KCNG2. If the concentration update count CFGCNG exceeds the reference count KCNG1, it is expected that the purge of the evaporated fuel from the canister 18 proceeds and the fuel concentration of the purge gas decreases. Therefore, if the concentration update count CFGCNG exceeds the reference count KCNG1, the process of step S558 is performed, and the value of the inflated value ΔDL is set to a predetermined value KΔDL2. The predetermined value KΔDL2 is set to a value equal to or greater than the upper limit value KAFFH of the dead band as described above.

  By the way, in a situation where the fuel concentration of the purge gas is expected to decrease, the vapor concentration correction coefficient FGPG, which is the learning value, also needs to be quickly changed in the low concentration direction, that is, in the positive direction. Therefore, in the present embodiment, the predetermined value KΔDL2 is not a constant value, but is determined according to the vapor concentration correction coefficient FGPG as shown in FIG. According to the setting shown in FIG. 23, the predetermined value KΔDL2 becomes a larger positive value as the vapor concentration correction coefficient FGPG is larger in the high concentration direction, that is, in the minus direction. With this setting, the learning update amount DLFGPG can be increased in the positive direction, that is, in the low concentration direction as the vapor concentration correction coefficient FGPG is in the high concentration direction. Learning delay in the low concentration direction can be avoided.

  If the density update count CFGCNG is equal to or smaller than the reference count KCNG1 as a result of the determination in step S548, the determinations in steps S550 and S552 are performed as processing for detecting the correction direction from the correction direction index smooth value AFFO. Here, the range of −KDFPG to + KDFPG is set as a dead band in which the padding correction by the padding value ΔDL is not performed. In step S550, it is determined whether or not the correction direction index smooth value AFFOSM is equal to or greater than the upper limit value of the dead band + KDFPG. In step S552, it is determined whether or not the correction direction index smooth value AFFOSM is less than or equal to the lower limit value −KDFPGKWP.

  As a result of the determination in steps S550 and S552, when the correction direction index smooth value AFFOSM is equal to or lower than the lower limit value −KDFPG of the dead band, it can be determined that the air-fuel ratio is shifted to the rich side. In this case, the process of step S556 is performed, and the value of the inflated value ΔDL is set to a predetermined value KΔDL1. The predetermined value KΔDL1 is set to a value equal to or greater than the upper limit value KAFFH of the dead band as described above. By adding the predetermined value KΔDL1 as the water increase value ΔDL, the value of the air-fuel ratio deviation index value AFF can be increased in the rich direction.

  On the other hand, as a result of the determination in steps S550 and S552, if the correction direction index smooth value AFFOSM is equal to or greater than the upper limit value of the dead band + KDFPG, it can be determined that the air-fuel ratio is shifted to the lean side. In this case, the process of step S558 is performed, and the value of the inflated value ΔDL is set to a predetermined value KΔDL2. By adding the predetermined value KΔDL2 as the inflated value ΔDL, the value of the air-fuel ratio deviation index value AFF can be inflated in the lean direction.

  If the correction direction index smooth value AFFOSM is within the dead band as a result of the determinations in steps S550 and S552, the direction of correction for increasing the air-fuel ratio deviation index value AFF is not determined. Therefore, the value of the inflated value ΔDL is set to the predetermined value KΔDL0, that is, zero by the processing in step S554.

  In step S560, the air-fuel ratio deviation index value AFF is calculated by the above-described equation (24) using the water addition value ΔDL set in any of steps S554, S556, and S558. The air-fuel ratio deviation index value AFF calculated in step S560 is taken into the routine shown in FIG. 2 and used for the calculation in step S2. In the determinations in steps S10 and S12, not the air-fuel ratio deviation index value AFF calculated by the equation (24) but the air-fuel ratio deviation index value AFF calculated by the equation (23), that is, the padding correction is performed. The previous air-fuel ratio deviation index value AFF is used.

  According to the routine shown in FIG. 22, in the situation where the temperature in the fuel tank 20 is high and the fuel concentration of the purge gas is unstable, the correction for increasing the air-fuel ratio deviation index value AFF is prohibited. In this case, since the learning update amount DLFGPG is calculated based on the air-fuel ratio deviation index value AFF that does not increase, it is possible to suppress the influence of the evaporation state of the fuel in the fuel tank 20 on the learning accuracy of the fuel concentration. .

  Further, according to the routine shown in FIG. 22, when the integrated purge amount from the start of the purge increases to a certain extent, the air-fuel ratio deviation index value AFF is increased and corrected in the lean direction by addition of the predetermined value KΔDL2. In this case, the learning update amount DLFGPG is calculated on the basis of the air-fuel ratio deviation index value AFF that is increased in the lean direction. Therefore, when the purge of evaporated fuel from the canister 18 proceeds and the fuel concentration of the purge gas decreases. The learning speed toward the low concentration side in the assumed situation can be increased, and the learning accuracy of the fuel concentration can be increased early.

  Further, according to the routine shown in FIG. 22, as in the routine shown in FIG. 21, when the air-fuel ratio is shifted to the lean side, the predetermined value KΔDL2 is added to the feedback correction coefficient FAF as the inflated value ΔDL. The learning update amount DLFGPG can be inflated in the low concentration direction. Conversely, when the air-fuel ratio is shifted to the rich side, the predetermined value KΔDL1 is added to the feedback correction coefficient FAF as the inflated value ΔDL, so that the learning update amount DLFGPG can be inflated in the high concentration direction. According to this, it is possible to increase the learning speed while providing the dead band, and it is possible to increase the learning accuracy of the vapor concentration correction coefficient FGPG at an early stage.

  Further, according to the routine shown in FIG. 22, the deviation ΔWP and the increase coefficient KRICH are also added to the feedback correction coefficient FAF as in the routine shown in FIG. 20 or FIG. The effect on the cost is offset. As a result, it is possible to obtain the learning update amount DLFGPG that reflects only the deviation of the air-fuel ratio due to the effect of the purge gas, and improves the learning accuracy of the vapor concentration correction coefficient FGPG in the situation where the adhesion state of the fuel adhering to the wall surface becomes unstable Can be made.

  In the present embodiment, the ECU 40 changes a part of the routine shown in FIG. 20 and executes the routine shown in FIG. 22, whereby the first, ninth, tenth, eleventh, twelfth and The “learning method correcting means” according to thirteenth invention is realized.

Embodiment 21. FIG.
Next, Embodiment 21 of the present invention will be described with reference to FIG. The system of the present embodiment is realized by causing the ECU 40 to execute a routine shown in FIG. 24 described later in place of the routine shown in FIG. 3 described above in the system of the first embodiment.

[Features of air-fuel ratio control according to Embodiment 21]
Also in the air-fuel ratio control according to the present embodiment, regarding the learning of the fuel concentration, correction control of the learning method is further performed as described below based on the concentration learning routine shown in the flowchart of FIG. The correction control according to the present embodiment calculates a correction value for canceling the influence of the change in the amount of fuel adhering to the wall on the feedback correction coefficient FAF, as in the correction control according to the eighteenth embodiment. A method of reflecting in the air-fuel ratio deviation index value AFF used in the concentration learning routine is adopted.

  In the correction control according to the present embodiment, the residual degree of the influence of the fuel cut is acquired as a parameter related to the stability of the fuel adhesion state, and the change in the amount of fuel adhering to the wall is based on this residual degree in the feedback correction coefficient FAF A correction value for canceling the influence is determined. The adhesion of fuel to the wall surface becomes more prominent as the amount of attached fuel when returning from the fuel cut is smaller, that is, the influence of the fuel cut remains. Therefore, by adding a correction value that increases according to the residual degree of influence to the air-fuel ratio deviation index value AFF, the influence of fuel adhesion on the wall surface upon return from the fuel cut has an effect on the air-fuel ratio. It can be reliably excluded from the value AFF.

  Further, in the correction control according to the present embodiment, the residual degree of the influence of the fuel increase is acquired as in the fifth embodiment, and the correction value is determined based on this residual degree. The fuel desorption from the wall surface becomes more prominent as the amount of fuel adhering to the wall surface is excessive, that is, as the influence of the fuel increase remains. Therefore, by adding a correction value that increases according to the residual degree of influence to the air-fuel ratio deviation index value AFF, the effect of fuel desorption from the wall surface upon return from fuel increase on the air-fuel ratio is affected. It can be reliably excluded from the index value AFF.

[Specific Processing in Embodiment 21]
In the present embodiment, the air-fuel ratio deviation index value AFF is corrected in order to eliminate the influence of the change in the wall surface attached fuel amount according to the flowchart of FIG. The flowchart of FIG. 24 shows a routine of influence deletion control as correction control of the learning method. The routine shown in FIG. 24 is executed by the ECU 40, which is a control device for the internal combustion engine 2, together with the routine shown in FIG.

  In the first step S570 of the routine shown in FIG. 24, it is determined whether or not a fuel cut (F / C) is being executed. If the result of determination is that fuel cut has not been performed, the value of the fuel injection counter CINJ is increased by a fixed value in step S574. On the other hand, when the fuel cut is being executed, the value of the fuel injection counter CINJ is decreased by a certain value in step S572. The fuel injection counter CINJ is provided with a limit value in each of the increasing direction and the decreasing direction. The limit value in the increasing direction is the maximum value M, and the limit value in the decreasing direction is the minimum value zero.

  The amount of fuel adhering to the wall surface after returning from the fuel cut decreases as the time during which the fuel cut has been executed increases, and increases with the elapsed time after the return. The process of step S572 is equivalent to counting the execution time of fuel cut, and the process of step S574 is equivalent to counting the elapsed time after returning. Therefore, the value of the fuel injection counter CINJ has a meaning as a parameter related to the amount of fuel adhering to the wall surface after returning from the fuel cut. Also, the smaller the fuel adhering to the wall surface after the return, the leaner the air-fuel ratio becomes with the fuel adhering to the wall surface, so the value of the fuel injection counter CINJ becomes the lean direction of the air-fuel ratio after returning from the fuel cut. It also has a meaning as a parameter indicating the degree of deviation.

  In the next step S576, a correction value ΔWPC to be reflected in the air-fuel ratio deviation index value AFF is determined based on the value of the fuel injection counter CINJ increased or decreased in step S572 or S574. Since the correction value ΔWPC is a correction value for canceling out the deviation of the air-fuel ratio in the lean direction after returning from the fuel cut, it is set to a negative value as shown in the figure. The correction value ΔWPC approaches zero as the value of the fuel injection counter CINJ increases, and is set to zero when the fuel injection counter CINJ is equal to or greater than a predetermined value. This is because as the value of the fuel injection counter CINJ is larger, the adhesion state of the fuel adhered to the wall surface becomes more stable and the deviation of the air-fuel ratio in the lean direction becomes smaller. Further, the temperature of the internal combustion engine 2 (cooling water temperature) is also considered in setting the correction value ΔWPC. The lower the temperature of the internal combustion engine 2, the more unstable the state of the fuel adhering to the wall surface. Therefore, even if the fuel injection counter CINJ is the same, the correction value ΔWPC is set to a large value in the negative direction if the temperature of the internal combustion engine 2 is low. The

  In the next step S578, first, the increase coefficient KRICHX is smoothed in the time direction by the above equation (17). Further, the deviation ΔRICH between the increase coefficient smoothing value RICHSM and the current increase coefficient KRICHX is calculated by the above equation (18). This deviation ΔRICH has a meaning as a parameter indicating the degree of deviation of the air-fuel ratio in the rich direction after returning from the fuel increase.

In step S580, the air-fuel ratio deviation index value AFF is calculated by the following equation (27) using the correction value ΔWPC obtained in step S576 and the deviation ΔRICH obtained in step S578. The air-fuel ratio deviation index value AFF calculated by the equation (27) is obtained by adding the correction value ΔWPC, the increase coefficient KRICH, and the deviation ΔRICH to the air-fuel ratio deviation index value AFF calculated by the equation (4). It corresponds to.
AFF = FAF + (ABYF / ABYF0) + ΔWPC + KRICH + ΔRICH (27)

  The air-fuel ratio deviation index value AFF calculated in the routine shown in FIG. 24 is taken into the routine shown in FIG. 2 and used for the calculation in step S2 and the determinations in steps S10 and S12.

  The feedback correction coefficient FAF reflects the deviation of the air-fuel ratio from the target air-fuel ratio, and in addition to the deviation of the air-fuel ratio due to the effect of the purge gas, the deviation of the air-fuel ratio caused by the change in the amount of fuel attached to the wall Is also included. In particular, after the return from the fuel cut or after the return from the fuel increase, the deviation of the air-fuel ratio due to the change in the amount of fuel attached to the wall surface becomes significant. According to the routine shown in FIG. 24, the correction value ΔWPC, the increase coefficient KRICH, and the deviation ΔRICH are added to the feedback correction coefficient FAF, thereby canceling the influence of factors other than the purge gas on the feedback correction coefficient FAF. . Therefore, the air-fuel ratio deviation index value AFF calculated by the equation (27) reflects only the air-fuel ratio deviation due to the influence of the purge gas, and the learning update amount DLFGPG is calculated using this air-fuel ratio deviation index value AFF. By doing so, it is possible to improve the learning accuracy of the vapor concentration correction coefficient FGPG in the situation where the adhesion state of the fuel adhering to the wall surface is particularly unstable after returning from the fuel cut or returning from the fuel increase. .

  In the present embodiment, the “learning method correcting means” according to the first, ninth, fourteenth, fifteenth and sixteenth inventions is realized by the ECU 40 executing the routine shown in FIG. Yes.

Others.
As mentioned above, although typical embodiment of this invention was described, this invention is not limited to said embodiment, It can implement in various deformation | transformation in the range which does not deviate from the meaning of this invention. . For example, the correction control according to each of the above embodiments may be combined appropriately.

1 is a schematic configuration diagram of an engine system to which an air-fuel ratio control device for an internal combustion engine according to a first embodiment of the present invention is applied. 3 is a flowchart showing a basic routine used for learning the fuel concentration in the first embodiment. 3 is a flowchart showing a dead zone expansion control routine executed in the first embodiment. 6 is a flowchart showing a dead zone expansion control routine executed in the second embodiment. 10 is a flowchart showing a dead zone expansion control routine executed in the third embodiment. 14 is a flowchart showing a dead zone expansion control routine executed in the fourth embodiment. 16 is a flowchart showing a dead zone expansion control routine executed in the fifth embodiment. 20 is a flowchart showing a routine of update reduction control performed in the sixth embodiment. 18 is a flowchart showing a routine of update reduction control performed in the seventh embodiment. FIG. 20 is a flowchart showing a routine of update reduction control performed in the eighth embodiment. FIG. 38 is a flowchart showing a routine of update cycle extension control performed in the ninth embodiment. FIG. 38 is a flowchart showing a routine of update cycle extension control performed in the tenth embodiment. FIG. 38 is a flowchart showing a routine for update cycle extension control performed in the eleventh embodiment. FIG. 38 is a flowchart showing a routine for update cycle extension control implemented in Embodiment 12. FIG. FIG. 38 is a flowchart showing an update stop control routine performed in the thirteenth embodiment. FIG. 25 is a flowchart illustrating a routine for update stop control performed in the fourteenth embodiment. FIG. 38 is a flowchart showing a routine that is executed for calculating a reference number KTWPWP in the fifteenth embodiment. FIG. 38 is a flowchart showing a routine that is executed for calculating a reference number KTWPWP in the sixteenth embodiment. FIG. 38 is a flowchart showing a routine that is executed for calculating a reference number KTWPWP in the seventeenth embodiment. FIG. 29 is a flowchart illustrating an influence deletion control routine performed in the eighteenth embodiment. FIG. 38 is a flowchart illustrating an influence deletion control routine performed in the nineteenth embodiment. FIG. 29 is a flowchart showing a routine for effect deletion control implemented in Embodiment 20. FIG. It is a figure which shows the example of a setting of positive direction correction value K (DELTA) DL2 used for calculation of smoothing correction value (DELTA) DL in the routine shown in FIG. FIG. 24 is a flowchart showing an influence deletion control routine performed in the twenty-first embodiment.

Explanation of symbols

2 Internal combustion engine 4 Combustion chamber 6 Intake passage 8 Exhaust passage 12 Fuel injection valve 16 Purge control valve 18 Canister 20 Fuel tank 30 Water temperature sensor 32 Rotational speed sensor 34 Exhaust gas sensor 36 Air flow meter 38 Acceleration position sensor 40 ECU (Electronic Control Unit)

Claims (13)

A fuel injection valve for injecting fuel into the intake passage of the internal combustion engine;
A canister that adsorbs the evaporated fuel generated in the fuel tank;
A purge mechanism for flowing purge gas containing evaporated fuel from the canister into the intake passage;
An exhaust gas sensor disposed in the exhaust passage of the internal combustion engine and outputting a signal corresponding to the air-fuel ratio of the exhaust gas;
A feedback correction coefficient for calculating a correction coefficient (hereinafter referred to as feedback correction coefficient) to be applied to the fuel injection amount in order to realize the target air-fuel ratio in accordance with the difference between the target air-fuel ratio and the air-fuel ratio corresponding to the exhaust gas sensor signal. A calculation means;
A concentration learning means for learning the fuel concentration of the purge gas based on a steady deviation from a reference value corresponding to the target air-fuel ratio of the feedback correction coefficient during operation of the purge mechanism;
Purge correction coefficient calculating means for calculating a correction coefficient (hereinafter referred to as a purge correction coefficient) to be applied to the fuel injection amount in order to cancel the influence of the purge gas on the air-fuel ratio based on the learned value of the fuel concentration;
Fuel injection amount calculating means for calculating the fuel injection amount of the fuel injection valve by reflecting the feedback correction coefficient and the purge correction coefficient in the basic fuel amount;
Learning method correction means for acquiring a parameter related to the stability of the adhesion state of the fuel adhering to the wall surface adhering to the intake passage, and correcting the fuel concentration learning method by the concentration learning means according to the value of the parameter ; With
The learning method correcting means includes
(1) Deviation between the ratio of the fuel injection amount to the total fuel amount used for combustion and the smooth value obtained by smoothing the ratio in the time direction,
(2) Cooling water temperature of the internal combustion engine,
(3) Elapsed time after returning from fuel cut,
(4) Deviation between an increase coefficient reflected in the basic fuel amount for increasing the fuel injection amount and a smooth value obtained by smoothing the increase coefficient in the time direction;
An air-fuel ratio control apparatus for an internal combustion engine, wherein any one of the parameters is acquired as the parameter . The concentration learning means is configured to update the learning value when a deviation of the feedback correction coefficient from the reference value exceeds a predetermined dead band,
2. The air-fuel ratio control apparatus for an internal combustion engine according to claim 1, wherein the learning method correcting means expands the dead band in accordance with low stability of the fuel adhesion state determined from the value of the parameter. .   The learning method correcting means changes either the upper limit value or the lower limit value of the dead band according to the direction of change in the amount of fuel adhering to the wall surface determined from the value of the parameter. An air-fuel ratio control apparatus for an internal combustion engine as described.   2. The internal combustion engine engine according to claim 1, wherein the learning method correcting unit reduces the update amount of the learning value in accordance with low stability of the fuel adhesion state determined from the parameter value. Fuel ratio control device. The concentration learning means is configured to learn a fuel concentration based on a smooth value obtained by smoothing a deviation of the feedback correction coefficient with respect to the reference value in a time direction,
2. The internal combustion engine according to claim 1, wherein the learning method correcting unit increases a smoothing coefficient used for smoothing according to a low stability of the fuel adhesion state determined from the value of the parameter. Air-fuel ratio control device.   2. The internal combustion engine engine according to claim 1, wherein the learning method correcting unit extends an update period of the learning value according to low stability of the fuel adhesion state determined from the value of the parameter. Fuel ratio control device.   2. The air-fuel ratio control of the internal combustion engine according to claim 1, wherein the learning method correcting means interrupts updating of the learning value when it is determined from the parameter value that the stability of the fuel adhesion state is low. apparatus.   8. The internal combustion engine according to claim 7, wherein the learning method correcting means restarts the updating of the learning value after a predetermined period of time when it is determined from the parameter value that the fuel adhesion state is stable. Air-fuel ratio control device.   The learning method correcting means calculates a correction value for canceling the influence of a change in the amount of fuel adhering to the wall surface on the feedback correction coefficient based on the value of the parameter, and the feedback correction coefficient used for learning the fuel concentration 2. The air-fuel ratio control apparatus for an internal combustion engine according to claim 1, wherein the correction value is reflected on the internal combustion engine. The concentration learning means is configured to update the learning value when a deviation of the feedback correction coefficient from the reference value exceeds a predetermined dead band,
When the deviation is positive, the learning method correction means reflects the added value set to be equal to or higher than the upper limit value of the dead band in the feedback correction coefficient used for learning the fuel concentration, and when the deviation is negative 10. The air-fuel ratio control apparatus for an internal combustion engine according to claim 9, wherein a subtraction value set to be equal to or lower than a lower limit value of the dead band is reflected in the feedback correction coefficient used for learning the fuel concentration.   11. The learning method correcting means implements reflection of the addition value or subtraction value in the feedback correction coefficient in a situation where it is determined that the amount of evaporated fuel in the fuel tank is small. An air-fuel ratio control apparatus for an internal combustion engine.   12. The learning method correction unit according to claim 10, wherein the added value is reflected in the feedback correction coefficient after the accumulated purge amount from the start of purge or a parameter related thereto reaches a predetermined value. An air-fuel ratio control apparatus for an internal combustion engine. A fuel injection valve for injecting fuel into the intake passage of the internal combustion engine;
A canister that adsorbs the evaporated fuel generated in the fuel tank;
A purge mechanism for flowing purge gas containing evaporated fuel from the canister into the intake passage;
An exhaust gas sensor disposed in the exhaust passage of the internal combustion engine and outputting a signal corresponding to the air-fuel ratio of the exhaust gas;
A feedback correction coefficient for calculating a correction coefficient (hereinafter referred to as feedback correction coefficient) to be applied to the fuel injection amount in order to realize the target air-fuel ratio in accordance with the difference between the target air-fuel ratio and the air-fuel ratio corresponding to the exhaust gas sensor signal. A calculation means;
A concentration learning means for learning the fuel concentration of the purge gas based on a steady deviation from a reference value corresponding to the target air-fuel ratio of the feedback correction coefficient during operation of the purge mechanism;
Purge correction coefficient calculating means for calculating a correction coefficient (hereinafter referred to as a purge correction coefficient) to be applied to the fuel injection amount in order to cancel the influence of the purge gas on the air-fuel ratio based on the learned value of the fuel concentration;
Fuel injection amount calculating means for calculating the fuel injection amount of the fuel injection valve by reflecting the feedback correction coefficient and the purge correction coefficient in the basic fuel amount;
Learning method correction means for acquiring a parameter related to the stability of the adhesion state of the fuel adhering to the wall surface adhering to the intake passage, and correcting the fuel concentration learning method by the concentration learning means according to the value of the parameter; With
The concentration learning means is configured to update the learning value when a deviation of the feedback correction coefficient from the reference value exceeds a predetermined dead band,
The learning method correcting means changes either the upper limit value or the lower limit value of the dead band to the side that expands the dead band according to the direction of change in the amount of fuel adhering to the wall surface determined from the value of the parameter. An air-fuel ratio control apparatus for an internal combustion engine characterized by the above.

Images