JP3303732B2 - Control device for internal combustion engine - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a control device for an internal combustion engine, and more particularly to a control device for an internal combustion engine that controls an air-fuel ratio to be leaner than a stoichiometric air-fuel ratio at least in a predetermined operating state.

[0002]

2. Description of the Related Art In order to reduce the fuel consumption rate, there is known a lean burn internal combustion engine which burns at an air-fuel ratio leaner than a stoichiometric air-fuel ratio at least in a predetermined operation state. In lean combustion, the less the more NO _X generation amount to the air-fuel ratio to lean, in order to be implemented further reduction in fuel consumption rate, it is preferable that the air-fuel ratio lean to the extent that combustion is not deteriorated.

[0003] Accordingly, in a lean burn internal combustion engine,
Lean limit control that monitors the amount of torque fluctuation in each cylinder and gently controls the air-fuel ratio so that the average value of the amount of torque fluctuation in each cylinder becomes the target amount of torque fluctuation, thereby making the air-fuel ratio as lean as possible Has been proposed. In such lean limit control, it is necessary to grasp an accurate torque fluctuation amount for each cylinder. For this purpose, a combustion pressure sensor may be provided for each cylinder, but this considerably increases the cost.

Japanese Patent Publication No. 7-33809 discloses an accurate torque for each cylinder utilizing a crank angle sensor required for general control of an internal combustion engine without using an expensive combustion pressure sensor. It is disclosed that the fluctuation is calculated and the above-described lean limit fuel injection amount control is performed.

[0005]

According to the 0005] The foregoing prior art, if the normal time, fuel consumption and NO _X generation amount of reduced lean combustion by a large margin, the use of expensive combustion pressure sensor It can be realized without. However, in lean combustion, misfires are likely to occur. The control cannot not only suppress such misfires, but also may lead to severe misfires in which no combustion occurs.

SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to provide a control device for an internal combustion engine which can reliably detect misfire by using a crank angle sensor and can surely suppress the misfire.

[0007]

[0008]

According to a first aspect of the present invention, there is provided a control apparatus for an internal combustion engine for controlling an air-fuel ratio to be leaner than a stoichiometric air-fuel ratio in at least a predetermined operating state. The first crank angle range is set within the crank angle range from the end of the stroke to the beginning of the explosion stroke, and the second crank angle is set within the crank angle range of the middle of the explosion stroke separated by a certain crank angle from the first crank angle range. Setting a crank angle range to detect a first angular velocity of the crankshaft within the first crank angle range and detecting a second angular velocity of the crankshaft within the second crank angle range; A driving force variation calculating means for calculating a driving force variation generated by each cylinder based on the angular velocity of the cylinder and the second angular velocity; First air-fuel ratio control means for controlling the air-fuel ratio so that the average value of the driving force fluctuation amount of each cylinder calculated by the amount calculating means becomes the target driving force fluctuation amount, and calculation by the driving force fluctuation amount calculating means A misfire level judging means for judging a misfire level in accordance with the magnitude of the driving force variation of at least one of the cylinders, and a misfire level determined by the misfire level judging means as a first level. And a second air-fuel ratio control means for stopping the air-fuel ratio control by the first air-fuel ratio control means and controlling the air-fuel ratio in accordance with the misfire level.

According to a second aspect of the present invention, in the control device for an internal combustion engine according to the first aspect, the misfire level determined by the misfire level determining means is equal to the first misfire level. When the misfire level exceeds the second level, the air-fuel ratio is leaner than the air-fuel ratio when the misfire level exceeds the second level in the air-fuel ratio control by the first air-fuel ratio control means or the second air-fuel ratio control means. It is further characterized by further comprising a guard means for guarding against occurrence.

According to a third aspect of the present invention, there is provided a control apparatus for an internal combustion engine for controlling an air-fuel ratio to be leaner than a stoichiometric air-fuel ratio at least in a predetermined operating state. The first crank angle range is set within the crank angle range up to the beginning of the stroke, and the second crank angle range is set within the mid-explosion stroke crank angle range separated by a certain crank angle from the first crank angle range.
And detecting a first angular velocity of the crankshaft within the first crank angle range and detecting a second angular velocity of the crankshaft within the second crank angle range. Driving force variation calculating means for calculating a driving force variation generated by each cylinder based on the first angular velocity and the second angular velocity; and the driving force variation of each cylinder calculated by the driving force variation calculating means Air-fuel ratio control means for controlling the air-fuel ratio so that the average value of the target driving force fluctuation amount, and ignition timing control means for controlling the ignition timing so as to be the optimal ignition timing calculated based on the engine operating state, A misfire level determining means for determining a misfire level according to the magnitude of the driving force fluctuation amount of at least one cylinder calculated by the driving force fluctuation amount calculating means; Depending on the misfire level determined by Bell determining means characterized by comprising an ignition timing correction means for correcting the optimum ignition timing.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Referring to FIG. 1, reference numeral 1 denotes an engine main body having four cylinders including a first cylinder # 1, a second cylinder # 2, a third cylinder # 3, and a fourth cylinder # 4. . Each cylinder #
1, # 2, # 3, and # 4 are connected to the surge tank 3 via the corresponding intake branch pipes 2, and each fuel branch pipe 2 has a fuel that injects fuel toward the corresponding intake port. The injection valve 4 is attached. The surge tank 3 is connected to an air cleaner 7 via an intake duct 5 and an air flow meter 6, and a throttle valve 8 is arranged in the intake duct 5. On the other hand, the cylinders # 1, # 2, # 3, # 4 is connected to the casing 12 with a built-in _the NO _x absorbent 11 via an exhaust manifold 9 and an exhaust pipe 10. This _the NO _x absorbent 11 is N the air-fuel ratio in the exhaust gas when the lean
It has a function of absorbing O _x and releasing and reducing the absorbed NO _x when the air-fuel ratio becomes stoichiometric or rich.

The electronic control unit 20 is composed of a digital computer, and is connected to a ROM (read only memory) 22, a RAM (random access memory) 23, a CPU (microprocessor) 24, and a power supply connected to each other by a bidirectional bus 21. Backup RAM connected
25, an input port 26 and an output port 27. A rotor 14 with external teeth is attached to the output shaft 13 of the engine, and a crank angle sensor 15 composed of an electromagnetic pickup is arranged facing the external teeth of the rotor 14. In the embodiment shown in FIG. 1, external teeth are formed on the outer periphery of the rotor 14 at every 30 ° crank angle. For example, some external teeth are deleted in order to detect the compression top dead center of the first cylinder. ing. Therefore, the crank angle sensor 15 generates an output pulse every time the output shaft 13 rotates by 30 ° crank angle except for the portion where the external teeth are deleted, that is, the missing tooth portion, and the output pulse is input to the input port 26. You.

The air flow meter 6 generates an output voltage proportional to the amount of intake air, and the output voltage is input to an input port 26 via a corresponding AD converter 28. An idle switch 16 for detecting that the throttle valve 8 is at an idling opening is attached to the throttle valve 8, and an output signal of the idle switch 16 is input to an input port 26. An air-fuel ratio sensor (O ₂ sensor) 1 for detecting an air-fuel ratio is provided in the exhaust manifold 9.
The output signal of the air-fuel ratio sensor 17 is input to the input port 26 via the corresponding AD converter 28. On the other hand, the output port 27 is connected to the corresponding drive circuit 2
9 is connected to each fuel injection valve 4. Reference numeral 40 denotes a spark plug arranged in each cylinder. The output port 27 is connected to each ignition plug 40 via the corresponding drive circuit 30, and the ignition timing is controlled.

In the internal combustion engine shown in FIG. 1, the fuel injection time TA
U is calculated based on the following equation. TAU = TP ・ FLEAN ・ FLLFB ・ FAF + TA
UV Here, TP indicates a basic fuel injection time, FLEAN indicates a lean correction coefficient, FLLFB indicates a lean limit feedback correction coefficient, FAF indicates a stoichiometric air-fuel ratio feedback correction coefficient, and TAUV indicates an invalid injection time.

The basic fuel injection time TP indicates an injection time required to make the air-fuel ratio a stoichiometric air-fuel ratio. The basic fuel injection time TP is obtained by an experiment. The basic fuel injection time TP is a function of the engine load Q / N (intake air amount Q / engine speed N) and the engine speed N in the form of the map shown in FIG. Is stored in the ROM 22 in advance. The lean correction coefficient FLEAN is a correction coefficient for making the air-fuel ratio a lean air-fuel ratio. The lean correction coefficient FLEAN is stored in the ROM 22 in advance as a function of the engine load Q / N and the engine speed N in the form of a map shown in FIG. Is stored in

The lean limit feedback correction coefficient F
LLFB is a correction coefficient for maintaining the air-fuel ratio at the lean limit. In the embodiment according to the present invention, the learning region for the lean air-fuel ratio feedback control is divided into, for example, nine regions with respect to the intake air amount Q and the engine speed N as shown in FIG. Lean limit feedback correction coefficients FLLFFB _{11 to} FLLFB, respectively
₃₃ is set.

The stoichiometric air-fuel ratio feedback correction coefficient FAF
Is a coefficient for maintaining the air-fuel ratio at the stoichiometric air-fuel ratio. The stoichiometric air-fuel ratio feedback correction coefficient FAF is controlled based on the output signal of the air-fuel ratio sensor 17 when the air-fuel ratio is to be maintained at the stoichiometric air-fuel ratio. Move up and down.

As shown in FIG. 4, a lean correction coefficient FLEAN is determined in accordance with the operating state of the engine in an operating region surrounded by a broken line, and the air-fuel ratio is maintained at a lean air-fuel ratio in this operating region. Is done. On the other hand, the air-fuel ratio is maintained at the stoichiometric air-fuel ratio in the operation region outside the region surrounded by the broken line in FIG. When the air-fuel ratio is to be maintained at the stoichiometric air-fuel ratio, the lean correction coefficient FLEAN and the lean limit feedback correction coefficient FLLFB are fixed to 1.0, and the stoichiometric air-fuel ratio feedback correction coefficient FAF is controlled based on the output signal of the air-fuel ratio sensor 17. You.

On the other hand, when the air-fuel ratio is to be maintained at the lean air-fuel ratio, the stoichiometric air-fuel ratio feedback correction coefficient FAF is fixed at 1.0, that is, the feedback control based on the output signal of the air-fuel ratio sensor 17 is stopped, and the lean correction coefficient FLEAN and lean limit feedback correction coefficient F
The air-fuel ratio is controlled to the lean-limit air-fuel ratio by the LLFB.

Next, the lean limit feedback control will be described with reference to FIG. Figure 3 shows the relationship between the engine output torque variation amount and NO _x generation amount and the air-fuel ratio. Fuel consumption rate as the air-fuel ratio is lean becomes small, and the generation amount of the more NO _x air-fuel ratio becomes lean is reduced. Therefore, from these points, it is preferable that the air-fuel ratio be as lean as possible. However, when the air-fuel ratio becomes lean to a certain degree or more, combustion becomes unstable, and as a result, the amount of torque fluctuation increases as shown in FIG. Therefore, in the embodiment according to the present invention, as shown in FIG. 3, the air-fuel ratio is maintained within the air-fuel ratio control region where the torque fluctuation starts to increase.

That is, specifically, the lean correction coefficient FLE
AN is the lean limit feedback correction coefficient FLLF
When B is set to FLLFFB = 1.0, the air-fuel ratio is determined to be at the center of the air-fuel ratio control region shown in FIG. On the other hand, the lean limit feedback correction coefficient F
LLFB is controlled in the torque fluctuation control region shown in FIG. 3 in accordance with the torque fluctuation amount, and when the torque fluctuation amount becomes large, the lean limit feedback correction coefficient FLL
When FB is increased, that is, the air-fuel ratio is reduced, and the torque fluctuation amount is reduced, the lean limit feedback correction coefficient FLLFB is reduced, that is, the air-fuel ratio is increased. Thus, the air-fuel ratio is controlled within the air-fuel ratio control region shown in FIG.

Note that the lean limit feedback correction coefficient FLLFB is set so as to cover the engine operating region in which the lean correction coefficient FLEAN is determined. Can amount of torque fluctuation is reduced by a large margin the generation amount of the fuel consumption rate and NO _x while ensuring the operation of the good vehicle when it is controlled to within the torque fluctuation control region shown in FIG. However, in order to control the amount of torque fluctuation within the torque fluctuation control region, the amount of torque fluctuation must be detected.

By the way, various methods have been conventionally proposed for calculating the amount of torque fluctuation. A typical example is a method of mounting a combustion pressure sensor in a combustion chamber and calculating a torque fluctuation amount based on an output signal of the combustion pressure sensor,
The amount of torque fluctuation based on the difference between the square of the first angular velocity ωa in the crank angle region from the end of the compression stroke to the early stage of the explosion stroke and the square of the second angular velocity ωb in the crank angle region in the middle of the explosion stroke. Is calculated.

The use of a combustion pressure sensor has the advantage that the torque generated by the cylinder to which the combustion pressure sensor is attached can be reliably detected, but has the disadvantage that a combustion pressure sensor is required. On the other hand, the angular velocities ωa, ω
b can be calculated from the output signal of the crank angle sensor provided in the internal combustion engine, so that the angular velocities ωa, ω
When the output torque is calculated based on b, there is an advantage that it is not necessary to provide a new sensor. However, in this case, if the engine drive system generates torsional vibration, there is a problem that the amount of torque fluctuation cannot be accurately detected. However, it is clear that a torque calculation method based on angular velocity that does not require a new sensor is preferable as long as this problem is solved. Therefore, in the present embodiment, the generated torque is calculated based on the angular velocity, and at this time, even if the engine drive system generates torsional vibration, the amount of torque fluctuation can be accurately detected.

Next, a method for calculating the output fluctuation and the torque fluctuation of the engine will be described. First, FIG. 6 shows a time of steady operation in which the engine drive system does not generate torsional vibration.
A method of calculating the driving force generated by each cylinder and the torque generated by each cylinder will be described with reference to (A) and (B). As described above, the crank angle sensor 15 generates an output pulse every time the crankshaft rotates by 30 ° crank angle.
It is arranged to generate an output pulse at the compression top dead center TDC of # 2, # 3 and # 4. Therefore, the crank angle sensor 15 generates an output pulse every 30 ° crank angle from the compression top dead center TDC of each of the cylinders # 1, # 2, # 3, and # 4. The ignition sequence of the internal combustion engine used in the present embodiment is 1-3-4-2.

6 (A) and 6 (B), the vertical axis T30 represents the elapsed time of 30 ° crank angle from the time when the crank angle sensor 15 generates an output pulse to the time when the next output pulse is generated. Further, Ta (i) indicates the elapsed time from the compression top dead center (hereinafter, referred to as TDC) of the i-th cylinder to 30 ° after the compression top dead center (hereinafter, referred to as ATDC),
Tb (i) is from ATDC60 ° of the i-th cylinder to ATDC9
The elapsed time up to 0 ° is shown. Therefore, for example, Ta
(1) indicates the elapsed time from TDC of the first cylinder to 30 ° ATDC, and Tb (1) indicates the ATDC of the first cylinder.
This indicates the elapsed time from 60 ° to ATDC 90 °. On the other hand, the 30 ° crank angle is changed to the elapsed time T3
When divided by 0, the result of the division represents the angular velocity ω. In the embodiment according to the present invention, 30 ° crank angle / Ta
(I) is referred to as a first angular velocity ωa in the i-th cylinder, and 3
0 ° crank angle / Tb (i) is set to the second
Is referred to as the angular velocity ωb. Therefore, 30 ° crank angle / T
a (1) represents the first angular velocity ωa of the first cylinder,
° Crank angle / Tb (1) represents the second angular velocity ωb of the first cylinder.

Looking at the first cylinder in FIGS. 6A and 6B, when the combustion starts and the combustion pressure increases, the elapsed time decreases from Ta (1) to Tb (1), and then. Tb
It rises again from (1). In other words, the angular velocity ω of the crankshaft is changed from the first angular velocity ωa to the second angular velocity ωb
And then fall again from the second angular velocity ωb. That is, the angular velocity ω of the crankshaft is determined by the combustion pressure.
Has been increased from the first angular velocity ωa to the second angular velocity ωb. FIG. 6A shows a case where the combustion pressure is relatively high, and FIG. 6B shows a case where the combustion pressure is relatively low. 6A and 6B, when the combustion pressure is high, the amount of decrease in the elapsed time (Ta (i) −Tb (i)) is greater than when the combustion pressure is low, and therefore the angular velocity ω is increased. The large amount (ωb−ωa) increases. When the combustion pressure increases, the driving force generated by the cylinder increases. Therefore, when the increase amount (ωb−ωa) of the angular velocity ω increases, the driving force generated by the cylinder increases. Therefore the first
(Ωb−ωa) between the second angular velocity ωb and the second angular velocity ωa
, The driving force generated by the cylinder can be calculated.

On the other hand, assuming that the rotational inertia moment of the engine is I, the kinetic energy is (1/2) Iωa ^{2 due} to the combustion pressure.
From () Iωb ² . Increased amount of kinetic energy (1/2) · I · (ωb 2 -ωa 2)
It is capable of calculating the square and the torque generated by the cylinder from the difference (ωb ² -ωa ²⁾ of the square of the second angular velocity [omega] b for represents the torque that cylinder will occur, thus the first angular velocity ωa Become.

By detecting the first angular velocity ωa and the second angular velocity ωb as described above, the driving force generated by the corresponding cylinder and the torque generated by the corresponding cylinder can be calculated from these detected values. The changes in the elapsed time T30 shown in FIGS. 6A and 6B slightly differ depending on the engine. Therefore, the crank angle range for detecting the first angular velocity ωa and the crank angle for detecting the second angular velocity ωb The range is set so that (ωb−ωa) best represents the driving force generated by the engine, or (ωb ² −ω) depending on the engine.
a ² ) is determined to best represent the torque generated by the engine. Therefore, depending on the engine, the crank angle range in which the first angular velocity ωa should be detected is BTDC3 before compression top dead center.
0 ° to TDC, and the crank angle range in which the second angular velocity ωb is to be detected is from ATDC90 ° to ATDC120.
°.

Therefore, in general terms, how to detect each of the angular velocities ωa and ωb is as follows. The first crank angle range is set in the crank angle range from the end of the compression stroke to the beginning of the explosion stroke. A second crank angle range is set in a crank angle region in a middle stage of an explosion stroke separated by a certain crank angle from the first crank angle range, a first angular velocity ωa of the crankshaft in the first crank angle range is detected, This means that the second angular velocity ωb of the crankshaft within the crank angle range is detected.

If the angular velocities ωa and ωb are detected as described above, the driving force and torque generated by the corresponding cylinder can be calculated based on the detected values. However, in the engine drive system, torsional vibration oscillating at the natural frequency of the drive system is generated by the explosion effect sequentially performed in each cylinder, and when the torsional vibration is generated in the engine drive system, the angular velocity ωa , Ωb, the driving force and torque generated by the cylinder cannot be accurately calculated. Next, this will be described with reference to FIGS.

FIG. 7 shows an elapsed time Ta sequentially calculated for each cylinder when torsional vibration occurs in the engine drive system.
The change of (i) is shown. When torsional vibration occurs in the engine drive system, the torsional vibration causes the angular velocity of the crankshaft to periodically increase and decrease, so that the elapsed time Ta
(I) periodically increases and decreases as shown in FIG.

FIG. 8 shows the elapsed time Ta in FIG.
The portion where (i) is decreasing is shown in an enlarged manner. FIG.
The elapsed time Ta (i) is equal to Ta (1) and T as shown in FIG.
a (3) decreases by ho time.
It is considered that the decrease in time is due to an increase in the amount of torsion due to torsional vibration. In this case, between Ta (1) and Ta (3), the amount of decrease in the elapsed time due to the torsional vibration is considered to increase almost linearly with the passage of time, and therefore, the amount of decrease in the elapsed time due to the torsional vibration. Are Ta (1) and Ta
It is represented by the difference between the dashed line connecting (3) and the horizontal line passing through Ta (1). Therefore, between Ta (1) and Tb (1), the elapsed time is reduced by h due to torsional vibration.

As described above, when torsional vibration occurs in the engine drive system, the elapsed time of Tb (1) decreases with respect to Ta (1). This includes the amount of decrease h of the elapsed time due to the vibration. Therefore, in order to obtain only the elapsed time Tb (1) reduced by the combustion pressure, h must be added to Tb (1). That is, even if the driving force or torque generated by each cylinder is obtained based on the detected elapsed times Ta (i) and Tb (i), the true driving force or torque cannot be obtained, and thus the true driving force or torque cannot be obtained. The engine output fluctuation or torque fluctuation cannot be determined.

Further, in a multi-cylinder internal combustion engine, in addition to such torsional vibration of the engine drive system, torsional vibration of the crankshaft itself is generated. Output fluctuation or torque fluctuation cannot be obtained. Next, this will be described with reference to FIG. In a multi-cylinder internal combustion engine, for example, a four-cylinder internal combustion engine as shown in FIG. 1, a large torsional vibration of the crankshaft itself occurs in the first and second cylinders. That is, a cylinder in which torsional vibration of the crankshaft 13 itself hardly occurs, for example, a third cylinder #
In FIG. 3, from Ta (3) to Tb (3) as shown in FIG.
, The elapsed time gradually decreases from Ta (1) to Tb (1) in the first cylinder # 1, and Tb (1) due to torsional vibration of the crankshaft itself. The elapsed time becomes longer. As a result, for the first cylinder # 1, the detected elapsed time Ta
Even if the driving force or torque generated by the first cylinder # 1 is obtained based on (1) and Tb (1), the true driving force or torque cannot be obtained, and thus the true engine output fluctuation or The torque fluctuation cannot be determined.

Therefore, in the present embodiment, even if torsional vibration of the engine drive system occurs, or even if torsional vibration of the crankshaft 13 itself occurs, a calculation method capable of calculating a true engine output fluctuation or torque fluctuation. Is adopted. Next, this calculation method will be described with reference to FIG. In FIG. 10, Ta (1) _j-1 and Tb (1) _j-1 are 1
The elapsed time of the first cylinder # 1 during the previous combustion is shown, and Ta (1) _j and Tb (1) _j represent the elapsed time of the first cylinder # 1 during the next combustion. On the other hand, Ta
(3) _j-1 represents the elapsed time at the time of the preceding combustion of the third cylinder # 3, which is performed subsequent to the combustion of the first cylinder # 1, and Ta (3) _j is the third cylinder # 3 represents the elapsed time at the time of the next combustion.

First, Ta (1)_jAnd Ta (1)_j-1
DTa (1) (= Ta (1) _j-Ta
(1)_j-1), And Ta (3)_jAnd Ta (3)_j-1When
Difference DTa (3) (= Ta (3)_j-Ta (3)_j-1)
And ask. Next, the output torque of the first cylinder # 1 during the preceding combustion
Assuming that the output torque during the next combustion is the same
ATDC at the time of the next combustion of the first cylinder # 1
The virtual elapsed time from 60 ° to 90 ° ATDC is T
b '(1)_jAnd Tb '(1)_jAnd Tb (1)
_j-1Difference Kb (1) (= Tb '(1)_j-Tb (1)
_j-1). These differences DTa (1) and Kb (1)
And DTa (3) as the height on a straight line
When corrected, the result is as shown in FIG.

As shown in FIG. 11, the elapsed time difference DTa
The elapsed time difference between (1) and DTa (3) is reduced by _I0 time. Even if torsional vibration occurs in the crankshaft itself, the effect of the torsional vibration does not appear in the elapsed time difference, and therefore, the decrease in the _I0 time of the elapsed time difference is due to the torsional vibration of the engine drive system. In this case, DTa
It is considered that the amount of decrease in the elapsed time difference between (1) and DTa (3) due to torsional vibration of the engine drive system increases almost linearly with time. Therefore, assuming that the output torque of the first cylinder # 1 during the previous combustion is the same as the output torque during the next combustion, the elapsed time difference DTa (1) and Kb
It is considered that the elapsed time difference decreases by I due to torsional vibration of the engine drive system between (1). Therefore, FIG.
As can be seen from the above, the elapsed time difference Kb (1) is expressed by the following equation, assuming that the output torque of the first cylinder # 1 during the previous combustion and the output torque during the next combustion are the same. Kb (1) = (2DTa (1) + DTa (3)) / 3

By the way, assuming that the output torque of the first cylinder # 1 during the previous combustion and the output torque during the next combustion are the same, the virtual elapsed time Tb '(1) of the first cylinder # 1
_j is represented by the following equation. Tb '(1) _j = Tb (1) _j-1 + Kb (1) Accordingly, when the output torque of the first cylinder # 1 during the previous combustion is assumed to be the same as the output torque during the next combustion, 1 Virtual output torque D during the next combustion of cylinder # 1
NS (1) is represented by the following equation. DNS (1) = {30 ° / Tb ′ (1) _j ｝ ² − {30 ° / Ta (1) _j ｝ ² = {30 ° / (Tb (1) _j−1 + Kb (1))} ² − ｛30 ° / Ta (1) _j ｝ ²

On the other hand, when the actual output torque is calculated using the actual elapsed time Tb (1) _j from ATDC60 ° to ATDC90 ° during the next combustion of the first cylinder # 1, this output torque DN (1) becomes It is expressed by the following equation. DN (1) = {30 ° / Tb (1) _j } ² − {30 ° /
Ta (1) _j ｝ ^{2 In} this case, the virtual output torque DNS of the first cylinder # 1
The difference between (1) and the actual output torque DN (1) represents the amount of torque fluctuation of the first cylinder # 1, and therefore, the first cylinder # 1
The torque fluctuation amount DLN (1) is expressed by the following equation. DLN (1) = DNS (1) −DN (1) = {30 ° /
Tb ′ (1) _j ｝ ² − {30 ° / Tb (1) _j ｝ ² Generally, the torque fluctuation amount DLN (i) of the i-th cylinder can be expressed by the following equation. DLN (i) = DNS (i) −DN (i) = {30 ° /
Tb ′ (i)｝ ² − {30 ° / Tb (i)} ²

That is, the torque fluctuation amount of the i-th cylinder is equal to the square of the virtual second angular velocity of the i-th cylinder and the actual second angular velocity.
It is obtained from the deviation from the power. By calculating the amount of torque fluctuation by such a method, it is possible to accurately detect the amount of torque fluctuation of each cylinder even if torsional vibration occurs in the engine drive system or even if torsional vibration occurs in the crankshaft itself. Become.

On the other hand, when calculating the torque fluctuation amount by such a method, if there is a variation in the interval between the external teeth formed along the outer periphery of the rotor 14 (FIG. 1), the torque fluctuation amount of each cylinder is calculated. It cannot be detected accurately. Therefore, in the embodiment according to the present invention, even if the interval between the external teeth of the rotor 14 varies, the elapsed times Ta (i), Tb detected so as to accurately detect the torque fluctuation amount of each cylinder.
(I) is corrected. Next, the elapsed time Ta
The correction method (i) will be described with reference to FIG.

In the embodiment according to the present invention, the elapsed time Ta
(I) is calculated based on the following equation. Ta (i) = (time required from TDC of cylinder i to 30 ° ATDC) · (1 + KTa (i)) where KTa (i) represents a correction coefficient for the i-th cylinder, and these correction coefficients KTa ( i) is calculated when fuel supply is stopped during deceleration operation in which no torsional vibration of the engine drive system occurs and no torsional vibration of the crankshaft itself occurs.

FIG. 12 shows the elapsed time Ta during the deceleration operation.
The change of (i) is shown. Since the rotation speed of the crankshaft gradually decreases during the deceleration operation, the elapsed time Ta
(I) is Ta (1) _j−1 , Ta (3) _j−1 , Ta (4)
_j-1 , Ta (2) _j-1 , Ta (1) _j , Ta (3) _j ,
It gradually increases as represented by Ta (4) _j and Ta (2) _j . At this time, if the intervals between the external teeth of the rotor 14 are the same, each elapsed time Ta (i) is considered to change along the solid line in FIG. 12, and the detected elapsed time Ta
If (i) deviates from the solid line in FIG. 12, it is considered that the interval between the external teeth for detecting the elapsed time Ta (i) deviating from the solid line in FIG. 12 deviates from the regular interval. . Therefore, in the embodiment according to the present invention, the elapsed time Ta
When (i) deviates from the straight line shown in FIG. 12, the elapsed time Ta (i) is corrected by the correction coefficient KTa (i) so that the elapsed time Ta (i) is located on the solid line in FIG. Like that.

More specifically, in the embodiment according to the present invention, the average value TaAV _j-1 of the elapsed time Ta (i) of the four cylinders in the 720 ° crank angle range (j-1), (j).
(Ta (1) _j-1 + Ta (3) _j-1 + Ta (4) _j-1 +
Ta (2) _j-1 ) / 4}, TaAV _j } = (Ta (1)
_j + Ta (3) _j + Ta (4) _j + Ta (2) _j ) /
Are sequentially calculated. Next, each elapsed time Ta (4)
_j-1 , Ta (2) _j-1 , Ta (1) _j , Ta (3) _j are considered to have increased from the average value TaAV _{j -1} to the average value TaAV _j at an average constant increase rate; Elapsed time AAV (4), AAV (2), AAV for each cylinder when considered to have increased at an average constant increase rate
(1), AAV (3) is obtained from the following equation. AAV (4) = (TaAV _j −TaAV _j−1 ) · (1 /
8) + TaAV _j−1 AAV (2) = (TaAV _j −TaAV _j−1 ) · (3 /
8) + TaAV _j−1 AAV (1) = (TaAV _j −TaAV _j−1 ) · (5 /
8) + TaAV _j−1 AAV (3) = (TaAV _j −TaAV _j−1 ) · (7 /
8) + TaAV _j-1

Next, based on the following equation, these
Elapsed time AAV (4), AAV (2), AAV (1),
Elapsed time Ta actually detected for AAV (3)
(4) _j-1, Ta (2)_j-1, Ta (1)_j, Ta
(3)_jKAFC (4), KAFC (2),
KAFC (1) and KAFC (3) are obtained. KAFC (4) = ｛Ta (4)_j-1−AAV (4)｝ /
AAV (4) KAFC (2) = ｛Ta (2)_j-1−AAV (2)｝ /
AAV (2) KAFC (1) = ｛Ta (1)_j−AAV (1)｝ / A
AV (1) KAFC (3) = ｛Ta (3)_j−AAV (3)｝ / A
AV (3)

Next, the basic correction coefficient KTaB is calculated using the following equation.
(I) is obtained for each cylinder. KTaB (i) = KTa (i) _j-1 -KAFC (i) /
4 That is, the previous correction coefficient KTa (i) for each cylinder
_The basic correction coefficient KTaB (i) is calculated by subtracting 1/4 of the deviation ratio KAFC (i) from _j-1 .
Next, the average value tKTaM of the basic correction coefficient KTaB (i)
Is calculated based on the following equation. tKTaM = ｛KTaB (1) + KTaB (2) + KT
aB (3) + KTaB (4)｝ / 4

Next, as shown in the following equation, the correction coefficient KTa (i) for each cylinder is calculated by subtracting the average value tKTaM from each basic correction coefficient KTaB (i). KTa (i) = KTaB (i) -tKTaM

As described above, the correction coefficient KTa (i) _j-1
Is not directly corrected based on the deviation ratio KAFC (i), and the basic correction coefficient KTaB (i) is once obtained for the following reason. For example, there is a shift ratio KAFC (1) only for the first cylinder # 1, and this shift ratio KAF
The correction coefficient KTa of the first cylinder # 1 based on C (1)
It is assumed that only (1) is corrected. In this case, the first cylinder # 1
Only the elapsed time Ta (1) is increased or decreased. However, the elapsed time Ta of only the first cylinder # 1
When (1) is increased or decreased, the correction coefficients KTa (2) and KTa of the remaining cylinders # 2, # 3, and # 4 are changed.
(3), KTa (4) is shifted.

In order to prevent such a problem from occurring, a basic correction coefficient KTaB (i) is once obtained, and an average value tKT of the basic correction coefficient is calculated from the basic correction coefficient KTaB (i).
The final correction coefficient KTa is obtained by subtracting aM.
(I) is obtained. That is, for example, when the basic correction coefficient KTaB (1) of the first cylinder # 1 is increased, the basic correction coefficient KTaB (i) of all the cylinders # 1, # 2, # 3, and # 4 is calculated. Average value of tKTa
M is subtracted. Thus, when tKTaM is subtracted from the basic correction coefficient KTaB (i) of all cylinders, the correction coefficient KTa (1) of the first cylinder # 1 increases, but the correction coefficient KTa (1) of the remaining cylinders # 2, # 3, # 4 Correction coefficient KTa (2), KTa
(3), KTa (4) is reduced and KTa
(1), KTa (2), KTa (3) and KTa
The sum of (4) is always maintained at zero. Thus, KTa
(1), KTa (2), KTa (3) and KTa
Each correction coefficient KTa is such that the sum of (4) is always zero.
When (i) is corrected, one of the correction coefficients KTa
Even if (i) is corrected, other correction coefficients KTa (i) do not shift.

In the embodiment shown in FIG. 12, for example, the elapsed time AAV (1) when the elapsed time Ta (1) _{j of the} first cylinder # 1 is considered to have increased at an average constant increasing rate.
Is larger than the cylinder # 1, the deviation ratio K with respect to the first cylinder # 1.
AFC (1) has a positive value. As a result, the correction coefficient KTa
(1) is decreased until the deviation ratio KAFC (1) becomes zero, and when the deviation ratio KAFC (1) becomes zero, the value of the correction coefficient KTa (1) becomes a constant value.
At this time, the elapsed time Ta (1) matches the elapsed time AAV (1). Correction coefficient KTa (i) for all cylinders
Reaches a constant value, when the rotation speed of the rotor 14 is constant, the elapsed time Ta (i) of each cylinder corrected using the correction coefficient KTa (i) becomes equal. Therefore, even if the interval between the external teeth of the rotor 14 varies, the torque fluctuation of each cylinder can be accurately detected.

Returning again to the calculation of the torque fluctuation, when the torsional vibration occurs in the engine drive system, the elapsed time difference DTa
(I) varies as shown in FIG. However, the elapsed time difference DTa (i) fluctuates even when the vehicle travels on an uneven road, and at this time, the fluctuation range of DTa (i) may be extremely large. FIG. 13 shows the fluctuation of DTa (i) when the vehicle travels on an uneven road, and the AMP of FIG. 13 shows the minimum DTa (i) and the maximum DTa (i).
a (i), that is, the amplitude. This amplitude AM
When P is small, the output fluctuation and torque fluctuation of each cylinder can be accurately detected by the method described above.

However, when the amplitude AMP increases, D
The output fluctuation or torque fluctuation of the cylinder in which Ta (i) becomes maximum or minimum cannot be detected accurately. That is, assuming that, for example, in FIG. 13, for example, the first cylinder having the maximum DTa (i) is the first cylinder # 1, the decrease I due to the torsional vibration of the virtual elapsed time difference Kb (1) of the first cylinder # 1 is shown in FIG. 11 D
It is obtained from the inclination of the chain line connecting Ta (1) and DTa (3). However, near the time when DTa (1) becomes maximum, the amount of increase or decrease of the elapsed time due to torsional vibration is DTa (1).
(2), DTa (1), and a smooth curve passing through DTa (3). Therefore, when Kb (1) of the first cylinder # 1 is obtained from DTa (1) and DTa (3), Kb (1) is obtained. The value of 1) is calculated to be much smaller than the actual value. As a result, Kb (1) does not show a regular value, and thus the output fluctuation amount and the torque fluctuation amount cannot be accurately detected. When the amplitude AMP increases, the same occurs in a cylinder in which DTa (i) is minimized.

Further, the DT of the cylinder in which combustion was performed immediately before
Even in a cylinder in which DTa (i) suddenly changes with respect to a (i), the value of Kb (i) deviates from the actual value, so that output fluctuation and torque fluctuation cannot be detected accurately. Therefore, in the embodiment according to the present invention, when the amplitude AMP is large, D
The output fluctuation amount or the torque fluctuation amount is not obtained for the cylinder in which Ta (i) becomes the maximum or the minimum, and DTa (i) changes abruptly with respect to DTa (i) of the cylinder in which the combustion was performed immediately before. Also, the output fluctuation amount or the torque fluctuation amount is not calculated for the set cylinder.

Next, a routine for obtaining the torque fluctuation amount of each cylinder will be described with reference to FIGS. FIG. 25 shows the calculation timing of each value performed in each routine. FIG. 14 shows an interrupt routine performed every 30 ° crank angle. FIG.
First, the process proceeds to a routine (step 100) for calculating the elapsed time difference DTa (i) and the elapsed time Tb (i). This routine is shown in FIGS. Next, the routine proceeds to a routine (step 200) for checking whether calculation of torque fluctuation is permitted. This routine is shown in FIGS. Next, the routine proceeds to a routine (step 300) for calculating the torque fluctuation. This routine is shown in FIG. Next, the lean limit feedback correction coefficient F
The routine proceeds to a routine for calculating the average engine speed and the average intake air amount used for calculating LLFB. This routine is shown in FIG.
It is shown in FIG.

Elapsed time difference DTa (i) and elapsed time T
Referring to FIGS. 15 to 18 showing the calculation routine of b (i), first, at step 101, the time TIM
E is set to TIMEO. The electronic control unit 20 includes a free-run counter that indicates the time, and the time TIME is calculated from the count value of the free-run counter. Next, at step 102, the current time TIME is taken. Therefore, the TIMEO of step 101 is 30 °
This represents the time before the crank angle.

Next, at step 103, it is determined whether the i-th cylinder is at ATDC 30 ° or not. If the ATDC of the i-th cylinder is not 30 °, the process jumps to step 111 to determine whether or not the ATDC of the i-th cylinder is 90 °. If the current i-th cylinder is not at ATDC 90 °, the elapsed time difference DTa (i) and the elapsed time Tb
The calculation routine of (i) is completed.

On the other hand, if it is determined in step 103 that the ATDC of the i-th cylinder is 30 °, the routine proceeds to step 104, where the elapsed time Ta (i) calculated before the 720 ° crank angle is TaO (i). Is done. Next, at step 105, the TD of the i-th cylinder is calculated based on the following equation.
Final elapsed time Ta from C to 30 ° ATDC
(I) is calculated.

Ta (i) = (TIME-TIMEO).
(1 + KTa (i)) That is, for example, if it is the ATDC 30 ° of the first cylinder # 1 now, the final elapsed time Ta (1) from TDC of the first cylinder # 1 to ATDC 30 ° is (TIME−TIME)
O) · (1 + KTa (1)). Here, (TIME-TIMEO) represents the elapsed time Ta (1) actually measured by the crank angle sensor 15, and KT
a (1) is a correction coefficient for correcting an error due to the external tooth interval of the rotor 13, and accordingly, (TIME-TIME)
The final elapsed time Ta (1) obtained by multiplying (O) by (1 + KTa (1)) accurately represents the elapsed time during which the crankshaft rotates by 30 ° crank angle.

Next, at step 106, the elapsed time difference DTa (i) (= Ta (i) is obtained by subtracting the elapsed time TaO (i) calculated 720 ° crank angle before from the currently calculated elapsed time Ta (i). -TaO (i))
Is calculated. Next, at step 107, it is determined whether or not the ATDC of the second cylinder # 2 is 30 °. When the ATDC of the second cylinder # 2 is not 30 °, the process jumps to step 110, and the combustion is performed immediately before (i-
1) A flag XCAL (i-1) indicating that the torque variation of the cylinder No. should be calculated is set (XCAL (i-1) ←
1) is done. Next, the routine proceeds to step 111. In the embodiment according to the present invention, as described above, the ignition order is 1-3-4-2.
Therefore, assuming that the ATDC of the first cylinder # 1 is 30 ° at present, the flag XCAL (2) indicating that the torque fluctuation amount of the second cylinder # 2 in which combustion was performed immediately before should be calculated is set. . Similarly, as shown in FIG. 25, when the final elapsed time Ta (3) is calculated, the flag XCAL (1) is set, and when the final elapsed time Ta (4) is calculated, the flag XCAL (3) is calculated. Is set, and when the final elapsed time Ta (2) is calculated, the flag XCAL (4) is set.

On the other hand, if it is determined in step 111 that the ATDC of the i-th cylinder is 90 °, the process proceeds to step 112, where the elapsed time Tb (i) calculated before the 720 ° crank angle is set to TbO (i). . Next, at step 113, the ATDC6 of the i-th cylinder is determined based on the following equation.
Final elapsed time Tb from 0 ° to 90 ° ATDC
(I) is calculated.

Tb (i) = (TIME-TIMEO).
(1 + KTb (i)) That is, for example, if the ATDC of the first cylinder # 1 is currently 90 °, the ATDC of the first cylinder # 1 is 60 ° to ATDC9.
The final elapsed time Tb (1) up to 0 ° is (TIME-
TIMEO) · (1 + KTb (i)).
Also in this case, the value (1 + KTb (i)) for correcting the error due to the external tooth interval of the rotor 13 is (TIME-TIM).
EO), the final elapsed time Tb
(1) accurately represents the elapsed time during which the crankshaft rotates by 30 ° crank angle. Next, at step 114, the ATDC 90 of the current second cylinder # 2
° is determined. AT of current cylinder # 2
If DC is not 90 °, the calculation routine of the elapsed time difference DTa (i) and the elapsed time Tb (i) is completed.

On the other hand, if it is determined in step 107 that the ATDC of the second cylinder # 2 is currently 30 ° ATDC, the routine proceeds to step 108, where the elapsed time Ta (i) in the 720 ° crank angle range (j) of FIG. average TaAV _j of) is calculated. TaAV _j = (Ta (1) + Ta (3) + Ta (4) +
Ta (2)) / 4 where Ta (1), Ta (3), Ta (4), Ta
(2) is Ta (1) _j , Ta (3) _j ,
Ta (4) _j and Ta (2) _j respectively. Next, the routine proceeds to a routine for calculating the correction coefficient KTa (i) in step 109. This routine is shown in FIG.

Referring to FIG. 17, first, in step 1
In 20 whether or not the fuel supply is stopped during deceleration operation
Is determined. When fuel supply is not stopped during deceleration operation
Completes the processing cycle and stops fuel supply during deceleration operation
Sometimes the process proceeds to step 121. Step 121
Of the elapsed time Ta (i) calculated in step 108
Average value TaAV_jAnd 720 of FIG. 12 which has already been calculated.
° Elapsed time Ta in crank angle range (j-1)
Average value TaAV of (i)_j-1(= (Ta (1) + Ta
(3) + Ta (4) + Ta (2)) / 4) (where Ta
(1), Ta (3), Ta (4), Ta (2)
Ta (1) shown in 12_j-1, Ta (3) _j-1, Ta
(4)_j-1, Ta (2)_j-1Is equivalent to)
FIG. 12 when it is considered that the increase has occurred at a certain constant increase rate.
Elapsed time AAV (4), AAV (2), AAV
(1), AAV (3) is calculated as follows. AAV (4) = (TaAV_j-TaAV_j-1) ・ (1 /
8) + TaAV_j-1 AAV (2) = (TaAV_j-TaAV_j-1) ・ (3 /
8) + TaAV_j-1 AAV (1) = (TaAV_j-TaAV_j-1) ・ (5 /
8) + TaAV_j-1 AAV (3) = (TaAV_j-TaAV_j-1) ・ (7 /
8) + TaAV_j-1

Next, at step 122, these elapsed times AAV (4), AAV (2), AAV
(1) Elapsed time TaO (4), TaO (2), Ta (1), Ta actually detected for AAV (3)
(3) deviation ratio KAFC (4), KAFC (2), K
AFC (1) and KAFC (3) are calculated. Where T
aO (4), TaO (2), Ta (1) and Ta (3) are Ta (4) _j-1 and Ta shown in FIG. 12, respectively.
(2) _j-1 , Ta (1) _j , and Ta (3) _j . KAFC (4) = {TaO (4) -AAV (4)} / A
AV (4) KAFC (2) = {TaO (2) −AAV (2)} / A
AV (2) KAFC (1) = {Ta (1) -AAV (1)} / AA
V (1) KAFC (3) = {Ta (3) -AAV (3)} / AA
V (3)

Next, at step 123, the correction coefficient KTa
It is determined whether the learning condition of (i) is satisfied.
For example, when a certain period of time has elapsed since the fuel supply was stopped, the engine speed is within a predetermined range, and the vehicle is not traveling on a rough road, for example, the elapsed time difference DTa (i) shown in FIG. When the amplitude AMP does not exceed the set value, it is determined that the learning condition is satisfied. When the learning condition is not satisfied, the routine proceeds to step 128, where the integrated value KAFCI (i) of the deviation ratio KAFC (i) is made zero, and then at step 129, the integrated count value CK
AFC is cleared.

On the other hand, if the learning condition is satisfied, the routine proceeds to step 124, where the deviation ratio K for each cylinder is determined.
AFC (i) is added to the corresponding integrated value KAFCI (i), and then at step 125, the integrated count value C
KAFC is incremented by one. Next, at step 126, it is determined whether or not the integrated count value CKAFC has reached 8. When the accumulated count value CKAFC is not 8, the processing cycle is completed and the accumulated count value CK
When AFC reaches 8, the routine proceeds to step 127, where the correction coefficient K
Ta (i) is calculated. That is, when the deviation ratio KAFC (i) is integrated eight times for each cylinder, the routine proceeds to step 127, where the correction coefficient KTa (i) is calculated.

In step 127, the correction coefficient KTa (i) is calculated as follows. That is, first, each integrated value KAFCI (i) is replaced with KAFCE (i). Next, the basic correction coefficient KTaB (i) is calculated based on the following equation. KTaB (i) = KTa (i)-(KAFCI (i) /
8) / 4 That is, when the elapsed time Ta (i) actually detected is shifted from the elapsed time AAV (i) when it is considered that the elapsed time Ta (i) has increased at an average constant increase rate. Is the basic correction coefficient KTaB obtained by correcting the current correction coefficient KTa (i) by ４ of the average value of the integrated value KAFCI (i).
(I). Next, an average value tKTaM of the basic correction coefficient KTaB (i) for all cylinders is calculated based on the following equation. tKTaM = (KTaB (1) + KTaB (2) + KT
aB (3) + KTaB (4)) / 4

Next, the correction coefficient KTa (i) is updated by correcting the basic correction coefficient KTaB (i) with its average value tKTaM as shown in the following equation. KTa (i) = KTaB (i) -tKTaM In this way, the correction coefficient KTa (i) for each cylinder is updated when fuel injection is stopped during deceleration operation.

On the other hand, if it is determined in step 114 of FIG. 16 that the ATDC of the second cylinder # 2 is 90 ° at present, the routine proceeds to step 115, and the 720 ° crank angle is obtained by the same method as when TaAV _j is obtained based on the following equation. average TbAV _j of the elapsed time in the angular range Tb (i) is calculated. TbAV _j = (Tb (1) + Tb (3) + Tb (4) +
Tb (2)) / 4 Next, the routine proceeds to a routine for calculating the correction coefficient KTb (i) in step 116. This routine is shown in FIG.

Referring to FIG. 18, first, in step 1
At 30, it is determined whether or not the fuel supply is stopped during the deceleration operation. When the fuel supply is not stopped during the deceleration operation, the processing cycle is completed, and when the fuel supply is stopped during the deceleration operation, the process proceeds to step 131. The average value TbAV _j in step 131 the elapsed time Tb which is calculated in step 115 (i), the average value of the time elapsed in 720 ° crank angle range immediately preceding _{Tb (i) TbAV j-1} (=
(Tb (1) + Tb (3) + Tb (4) + Tb (2))
/ 4), the elapsed times BAV (4), BAV (2), and BAV when considered to increase at an average constant increase rate
AV (1) and BAV (3) are calculated. BAV (4) = (TbAV _j −TbAV _j−1 ) · (1 /
8) + TbAV _j−1 BAV (2) = (TbAV _j −TbAV _j−1 ) · (3 /
8) + TbAV _j−1 BAV (1) = (TbAV _j −TbAV _j−1 ) · (5 /
8) + TbAV _j−1 BAV (3) = (TbAV _j −TbAV _j−1 ) · (7 /
8) + TbAV _j-1

Next, at step 132, these elapsed times BAV (4), BAV (2), BAV
(1), elapsed times TbO (4), TbO (2), Tb (1), Tb actually detected for BAV (3)
(3) Deviation ratio KBFC (4), KBFC (2), K
BFC (1) and KBFC (3) are calculated. KBFC (4) = {TbO (4) -BAV (4)} / B
AV (4) KBFC (2) = {TbO (2) -BAV (2)} / B
AV (2) KBFC (1) = {Tb (1) -BAV (1)} / BA
V (1) KBFC (3) = {Tb (3) -BAV (3)} / BA
V (3)

Next, at step 133, the correction coefficient KTb
It is determined whether the learning condition of (i) is satisfied.
As described above, for example, a certain time has elapsed since the fuel supply was stopped, the engine speed is within a predetermined range,
When the vehicle is not traveling on a rough road, for example, when the amplitude AMP of the elapsed time difference DTa (i) shown in FIG. 13 does not exceed the set value, it is determined that the learning condition is satisfied. Step 13 when the learning condition is not satisfied
Proceeding to 8, the integrated value KBFCI of the deviation ratio KBFC (i)
(I) is set to zero, and then at step 139, the integrated count value CKBFC is cleared.

On the other hand, if the learning condition is satisfied, the routine proceeds to step 134, where the deviation ratio K for each cylinder is determined.
BFC (i) is added to the corresponding integrated value KBFCI (i), and then at step 135, the integrated count value C
KBFC is incremented by one. Next, at step 136, it is determined whether or not the integrated count value CKBFC has reached 8. When the accumulated count value CKBFC is not 8, the processing cycle is completed, and the accumulated count value CK
When BFC reaches 8, the routine proceeds to step 137, where the correction coefficient K
Tb (i) is calculated. That is, when the deviation ratio KBFC (i) is integrated eight times for each cylinder, the routine proceeds to step 137, where the correction coefficient KTb (i) is calculated.

In step 137, the correction coefficient KTb (i) is calculated as follows. That is, first, each integrated value KBFCI (i) is replaced with KBFCE (i). Next, the basic correction coefficient KTbB (i) is calculated based on the following equation. KTbB (i) = KTb (i)-(KBFCI (i) /
8) / 4 That is, when the actually detected elapsed time Tb (i) is shifted from the elapsed time BAV (i) when it is considered that the elapsed time Tb (i) increases at an average constant increase rate. Is the basic correction coefficient KTbB obtained by correcting the current correction coefficient KTb (i) by ４ of the average value of the integrated value KBFCI (i).
(I). Next, an average value tKTbM of the basic correction coefficients KTbB (i) for all cylinders is calculated based on the following equation. tKTbM = (KTbB (1) + KTbB (2) + KT
bB (3) + KTbB (4)) / 4

Next, the correction coefficient KTb (i) is updated by correcting the basic correction coefficient KTbB (i) with its average value tKTbM as shown in the following equation. KTb (i) = KTbB (i) -tKTbM In this manner, when the fuel injection is stopped during the deceleration operation, the correction coefficient KTb (i) for each cylinder is updated.

Next, the torque variation calculation permission check routine shown in FIGS. 19 to 21 will be described with reference to FIG. In this routine, the amplitude AMP of the variation of the elapsed time difference DTa (i) due to the vehicle traveling on the uneven road
This is provided to prohibit calculation of the torque fluctuation amount for a specific cylinder when (FIG. 13) becomes large.

That is, referring to FIGS. 19 to 21, first, at step 201, it is determined whether or not the ATDC of any of the cylinders is currently 30 °. If the ATDC of any of the cylinders is not 30 ° at present, the processing cycle is completed. If the ATDC of any of the cylinders is at 30 ° at present, the process proceeds to step 202. In steps 202 to 204, the maximum elapsed time difference DT30max when the elapsed time difference DTa (i) increases and then decreases is calculated. That is, in step 202, step 10 in FIG.
6, the DTa (i) calculated is the maximum elapsed time difference D
It is determined whether it is larger than T30max. DT3
If 0max> DTa (i), the routine jumps to step 205, whereas DT30max ≦ DTa (i)
In the case of, the routine proceeds to step 203, where DTa (i) is DT
30 max. Next, at step 204, DTa
An increase flag XMXRE indicating that (i) is increasing
C is set (XMXREC ← 1), and then the routine proceeds to step 205.

In steps 205 to 207, the minimum elapsed time difference DT30min when the elapsed time difference DTa (i) decreases and then increases is calculated. That is, in step 205, it is determined whether or not DTa (i) calculated in step 106 in FIG. 15 is smaller than the minimum elapsed time difference DT30min. DT30min <DTa
In the case of (i), the process jumps to step 208. On the other hand, when DT30min ≧ DTa (i), the process proceeds to step 206, where DTa (i) is set to DT30min. Next, at step 207, a decrease flag XMNREC indicating that DTa (i) is decreasing is set (XM
NREC ← 1), and then go to step 208.

In steps 208 to 214, D
The amplitude AMP of the fluctuation of Ta (i) (FIG. 13) is equal to the set value A _0.
Is exceeded, a prohibition flag is set to prohibit the calculation of the amount of torque fluctuation for the cylinder in which DTa (i) is maximized. That is, in step 208, DT30max>
It is determined whether DTa (i) and XMXREC = 1. DT30max ≦ DTa (i) or the increase flag XMXREC is reset (XMXREC =
If 0), jump to step 215,
On the other hand, if DT30max> DTa (i) and XM
When XREC = 1, the process proceeds to step 209.

[0081] That is, the No. 1 cylinder # 1 of the difference of elapsed times DTa at time t ₁ as shown in FIG. 22 (1) is to become the maximum. In this case, in the interruption routine performed at the time t ₁ proceeds from step 202 to step 203 DTa (1) is a DT30max, then the maximum flag XMXREC is set in step 204. On the other hand, in the interruption routine performed at the time t ₂ in FIG. 22 jumps from step 202 to step 205. At this time, in step 208, DT30ma
Since it is determined that x> DTa (3) and XMXREC = 1, the process proceeds to step 209. That is, the time t ₂ the flow proceeds to step 209 where the elapsed time DTa (i) starts to decrease.

At step 209, the maximum elapsed time difference DT3
0max is set to TMXREC. Then step 21
At 0, the minimum elapsed time difference from the maximum elapsed time difference TMXREC
TMNREC (determined in step 216 described later)
Is subtracted to obtain the amplitude AM of the variation of DTa (i).
P is calculated. Next, at step 211, when the minimum time elapses
The initial value of the difference DT30min is set to DTa (i).
Next, at step 212, the increase flag XMXREC is reset.
Set (XMXREC ← 0). Then step 2
At 13 the amplitude AMP is equal to the set value A₀Is greater than
Is determined. AMP <A₀At step 215
Jump. On the other hand, AMP ≧ A ₀When
Proceeding to step 214, the torque fluctuation calculation prohibition flag XNO
CAL is set (XNOCAL ← 1). That is,
22 time t_TwoIn the interrupt routine performed in
As described above, the torque fluctuation amount of the first cylinder # 1 is calculated.
You. Therefore, in this interrupt routine, AMP ≧ A₀When
And the torque fluctuation calculation prohibition flag XNOCAL is set.
Then, the calculation of the torque fluctuation amount of the first cylinder # 1, that is, D
Calculation of the torque fluctuation amount of the cylinder that maximizes Ta (i) is prohibited.
Is stopped.

In steps 215 to 221, D
Amplitude AMP of the variation of ta (i) is set prohibition flag for prohibiting calculation of the amount of torque fluctuation for cylinder DTa (i) is minimized when it exceeds the set value A _0.
That is, in step 215, DT30min <DTa
It is determined whether or not (i) and XMNREC = 1. If DT30min ≧ DTa (i) or the decrease flag XMNREC is reset (XMNREC = 0), the process jumps to step 222, whereas DT30min <DTa (i) and XMNRE
When C = 1, the process proceeds to step 216.

[0084] That is, the first cylinder # 1 of the difference of elapsed times DTa at time t ₃ as shown in FIG. 22 (1) is minimized. In this case, in the interruption routine performed at the time t ₃ proceeds from step 205 to step 206 DTa (1) is a DT30min, then decrease flag XMNREC is set in step 207. On the other hand, in the interruption routine performed at the time t ₄ in FIG. 22 jumps from step 205 to step 208. At this time, in step 215, DT30mi
Since it is determined that n <DTa (3) and XMNREC = 1, the process proceeds to step 216. That is, the time t ₄ when the elapsed time difference DTa proceed to step 216 (i) starts to increase.

In step 216, the minimum elapsed time difference DT3
0 min is set as TMNREC. Then step 21
In step 7, the amplitude AMP of the fluctuation of DTa (i) is calculated by subtracting the minimum elapsed time difference TMNREC from the maximum elapsed time difference TMXREC. Next, at step 218, the initial value of the maximum elapsed time difference DT30max is set to DTa.
(I). Next, at step 219, the decrease flag XMNREC is reset (XMNREC ← 0).
Then whether step 220 the amplitude AMP is greater than the set value A ₀ is judged. At the time of the AMP <A ₀ jumps to step 222. AMP ≧
Torque fluctuation calculation prohibition flag XNOCAL proceeds to step 221 when the A ₀ is set (XNOCAL ← 1)
Is done. That is, the torque variation of the first cylinder # 1 is in the interruption routine performed at the time t ₄ in FIG. 22 is calculated. Therefore calculation of AMP ≧ A ₀ becomes, when the torque fluctuation calculation prohibition flag XNOCAL is set 1 cylinder # 1 of the amount of torque fluctuation in this interrupt routine, i.e., D
Calculation of the torque fluctuation amount of the cylinder in which Ta (i) is minimized is prohibited.

In steps 222 and 223, the calculation of the torque fluctuation amount for the cylinder in which the elapsed time difference DTa (i) has suddenly changed is prohibited. That is, at step 222 | DTa (i-2) -DTa (i-1) | is K _o · |
DTa (i-1) -DTa (i) | is determined. Here, the constant _Ko is a value of about 3.0 to 4.0. In step 222, | DTa (i−
2) -DTa (i-1) | <K o · | DTa (i-1)
When it is determined that -DTa (i) |, the processing routine is completed, and | DTa (i-2) -DTa (i-1).
_{| ≧ K o · | DTa (} i-1) -DTa (i) | a is judged to be the torque fluctuation calculation prohibition flag XNOCAL proceeds to step 223 when the are set.

[0087] That is, when an interrupt routine at time t ₃ now Figure 22 at this time | DTa (4) -
DTa (2) | ≧ K _o · | DTa (2) −DTa (1)
| Is determined. As shown in FIG. 22, when DTa (2) changes rapidly with respect to DTa (4), | DT
a (4) -DTa (2) | is _{K o · | DTa (2)} -D
Ta (1) |. At this time, the torque fluctuation calculation prohibition flag is set, and the calculation of the torque fluctuation amount of the second cylinder # 2 in which the elapsed time difference DTa (i) has suddenly changed is prohibited.

Next, the torque fluctuation calculation routine shown in FIG. 23 will be described. Referring to FIG. 23, first, in step 301, combustion was performed immediately before (i-
1) It is determined whether or not a flag XCAL (i-1) indicating that the torque variation of the cylinder No. should be calculated is set. When the flag XCAL (i-1) = 0, that is, when the flag XCAL (i-1) is not set, the processing cycle is completed. On the other hand, the flag XCAL
When (i-1) = 1, that is, the flag XCAL (i-1)
Is set, the process proceeds to step 302, where the flag XCAL (i-1) is reset, and then proceeds to step 303.

In step 303, it is determined whether or not a prohibition flag XNOCAL for prohibiting the calculation of the amount of torque fluctuation for the cylinder in which combustion has been performed immediately before is reset (XNOCAL = 0). When this prohibition flag is set (XNOCAL = 1), step 31
Proceeding to 1, the prohibition flag XNOCAL is reset.
On the other hand, when the prohibition flag has been reset, the routine proceeds to step 304. That is, the process proceeds to step 304 only when the flag XCAL is set and the prohibition flag XNOCAL is reset.

In step 304, a virtual elapsed time difference Kb (i-1) of the (i-1) th cylinder in which combustion was performed immediately before.
(See FIGS. 10 and 11) is calculated based on the following equation. Kb (i−1) = ｛2 · DTa (i−1) + DTa
(I)｝ / 3 Next, at step 305, based on the following equation, the actual output torque DN of the (i-1) th cylinder in which combustion was performed immediately before:
(I-1) is calculated. DN (i−1) = {30 ° / Tb (i−1)} ² − ｛3
0 ° / Ta (i-1)｝ ²

Next, at step 306, the virtual output torque DNS (i-1) of the (i-1) th cylinder in which combustion was performed immediately before is calculated based on the following equation. DNS (i−1) = ｛30 ° / (TbO (i−1) + K
b (i-1)) ^2- {30 / Ta (i-1)} ² Then, in step 307, as shown in the following equation, the virtual output torque DNS (i-1) is converted to the actual output torque DN (i
By subtracting -1), the torque fluctuation amount DLN (i-1) of the (i-1) th cylinder in which combustion was performed immediately before is calculated. DLN (i-1) = DNS (i-1) -DN (i-1)

That is, for example, the ATDC3 of the third cylinder # 3
If it is 0 ° and the flag XCAL (1) is set, a virtual elapsed time difference Kb (1) with respect to the first cylinder # 1 is calculated in step 304, and then in step 305, the actual time difference Kb (1) of the first cylinder # 1 is calculated. Output torque DN
(1) is calculated, and in step 306, the first cylinder #
1 is calculated, and then in step 307, the torque fluctuation amount D of the first cylinder # 1 is calculated.
LN (1) is calculated.

The output fluctuation amount GLN (i-
When determining 1), steps 305 to 30
7, the following calculation may be performed. That is, in step 305, instead of obtaining the actual torque fluctuation amount DN (i-1), the actual output fluctuation amount GN (i-
1) is calculated. GN (i-1) = {30 ° / Tb (i-1)}-{30
° / Ta (i-1)｝ Next, at step 306, the virtual torque fluctuation amount DNS
Instead of obtaining (i-1), a virtual output fluctuation amount GNS (i-1) is calculated based on the following equation. GNS (i−1) = ｛30 ° / (TbO (i−1) + K
b (i-1))｝-{30 ° / Ta (i-1)}

Next, at step 307, the torque variation D
Instead of obtaining LN (i-1), output fluctuation amount GLN (i-1) is calculated based on the following equation. GLN (i-1) = GNS (i-1) -GN (i-1) In step 307, the torque fluctuation amount DLN (i
When -1) is calculated, the routine proceeds to step 308, where it is determined whether or not the torque fluctuation amount DLN (i-1) is positive.
If DLN (i−1) ≧ 0, the routine jumps to step 310 to change the torque fluctuation amount DL of the cylinder in which the combustion was performed immediately before.
Integration request flag X indicating that N (i-1) should be integrated
DLNI (i-1) is set (XDLNI (i-1) ←
1) is done. On the other hand, if DLN (i-1) <0, the routine proceeds to step 309, where DLN (i-1) is made zero, and then proceeds to step 310. Note that the torque of each cylinder repeatedly increases and decreases. In this case, the amount of torque fluctuation may be obtained by integrating either the increase in torque or the decrease in torque. In the routine shown in FIG. 23, only the amount of decrease in torque is integrated, and therefore, when DLN (i-1) <0, DLN (i-1) is set to zero as described above.

[0095] Next referring to FIG. 24 the engine speed average value N _AVE and the calculation of the intake air amount average value Q _{AV E} is described. The average engine speed N _AVE and the average intake air amount Q _AVE are used when calculating a lean limit feedback correction coefficient FLLFFB which will be described later. Referring to FIG. 24, first, in step 401, it is determined whether or not the ATDC of the third cylinder # 3 is currently 30 °. When the ATDC of the third cylinder # 3 is not 30 °, the processing cycle is completed.
When it is C30 °, the process proceeds to step 402. In step 402, it is determined whether a torque fluctuation value calculation condition for calculating a torque fluctuation value of the engine is satisfied.
For example, whether the condition to make the air-fuel ratio lean is not satisfied,
Alternatively, when the change amount ΔQ per unit time of the intake air amount Q is equal to or larger than a set value, or when the change amount ΔN per unit time of the engine speed is equal to or larger than the set value, the torque fluctuation value calculation condition is not satisfied. Otherwise, it is determined that the torque fluctuation value calculation condition is satisfied.

If it is determined in step 402 that the torque fluctuation value calculation condition is satisfied, step 40 is executed.
Proceeding to 9, the average value N _AVE of the engine speed and the average value Q _AVE of the intake air amount Q are calculated.

On the other hand, if it is determined in step 402 that the torque fluctuation value calculation condition is not satisfied, the routine proceeds to step 404, where the torque fluctuation amount DLN for each cylinder is determined.
The integrated value DLNI (i) of (i) (this integrated value is calculated in a routine described later) is cleared, and then at step 405, the integrated count value C for each cylinder is cleared.
DLNI (i) (this integrated count value is calculated in a routine described later) is cleared.

Next, at step 406, a target torque fluctuation value LVLLFB is calculated. In the embodiment according to the present invention, the air-fuel ratio is feedback-controlled such that the torque fluctuation value calculated as described later becomes the target torque fluctuation value LVLLFB. This target torque fluctuation value LVLLLF
B increases as the engine load Q / N increases, and increases as the engine speed N increases, as shown in FIG. The target torque fluctuation value LVLLFB is stored in the ROM 22 in advance in the form of a map as a function of the engine load Q / N and the engine speed N as shown in FIG. Next, at step 407, the torque fluctuation value DLNISM (i) of each cylinder
(This torque fluctuation value is calculated in a routine to be described later) is set as a target torque fluctuation value LVLLFB calculated from the map of FIG.

FIG. 28 shows a main routine that is repeatedly executed. In this main routine, first, a torque fluctuation value calculation and misfire level determination routine (step 600) is executed. This routine is shown in FIGS.
It is shown in FIG. Next, a routine for calculating the lean limit feedback correction coefficient FLLFB (step 70)
0) is executed. This routine is shown in FIGS. Next, when a predetermined crank angle is reached, an injection time calculation routine (step 800) is executed. This routine is shown in FIG. Next, another routine (step 900) is executed.

Next, the torque fluctuation value calculation and misfire level determination routine shown in FIGS. 29 to 31 will be described. First, at step 601, the torque fluctuation amount D
Integration request flag XD indicating that LN (i) should be integrated
It is determined whether LNI (i) is set (XDLNI (i) = 1). Integration request flag XDLNI
When (i) is not set, the processing cycle is completed, and when the integration request flag XDLNI (i) is set, the process proceeds to step 602. In step 602, the integration request flag XDLNI (i) is reset. Next, at step 603, the cylinder number for which the torque variation is currently integrated is stored as i. Next, at step 604, the torque variation DLN (i) is added to the torque variation integrated value DLNI (i). Next, at step 605, the integrated count value CDLNI (i) is incremented by one. That is, for example, step 601
, The integration request flag XDLNI for the first cylinder
If (1) is set, the flag XDLNI (1) is reset in step 602, the cylinder number 1 is stored as i in step 603, and the torque variation integrated value DLNI (1) is calculated in step 604. In 605, the accumulated count value CD
LNI (1) is incremented by one.

Next, at step 606, the cylinder number i
Is 1 or not. If the cylinder number is not 1, the process jumps to step 608. If the cylinder number is 1, the process proceeds to step 607. In step 607, the integrated count value CDLNI of the first cylinder
The maximum value of the torque fluctuation amount DLN (i) for each cylinder during the increment of (1) is stored as DLNMX (i). The maximum value DLNMX (i) of this torque fluctuation amount
Is used to determine the misfire level as will be described in detail later. For example, the maximum value of the torque fluctuation amount of each cylinder at the same time at the timing when the torque fluctuation amount of the first cylinder is integrated Is stored. Next, at step 608, the integrated count value CDL
It is determined whether NI (i) has become 16 or not. CDL
When NI (i) is not 16, the processing cycle is completed. When CDLNI (i) becomes 16, the routine proceeds to step 609, where the torque fluctuation value DLNISM of each cylinder is obtained from the following equation.
(I) is calculated. DLNISM (i) = DLNISM (i) + ｛DLNI
(I) -DLNISM (i)｝ / 4

Next, at step 610, the integrated torque variation DLNI (i) for each cylinder is cleared, and then at step 611, the integrated count value CDLNI (i).
Is reset.

That is, the calculated torque variation integrated value DL
NI (i) and torque fluctuation value DL used so far
When there is a difference from NISM (i), this difference ｛DL
NI (i) -DLNISM (i)} multiplied by 1/4 is added to torque fluctuation value DLNISM (i). Therefore, for example, the integrated count value CD for the first cylinder # 1
When LNI (1) becomes 16, the torque fluctuation value DLNISM (1) is calculated in step 609.

Next, at step 612, step 606 is executed.
It is determined whether or not the cylinder number i is 1 in the same manner as described above. If the cylinder number is not 1, the processing cycle is completed, but if the cylinder number is 1, step 6 is executed.
Proceeding to 13, the maximum value DLNM of the torque fluctuation amount for each cylinder
The maximum value of X (i) is stored as DLNMAX. That is, the torque fluctuation amount DLN (1) for the first cylinder
Is accumulated 16 times, the maximum torque variation generated in any one of the cylinders is stored as DLNMAX.

Next, at step 614, it is determined whether or not the maximum torque fluctuation amount DLNMAX is equal to or less than the set value LVLMF. When the maximum torque fluctuation amount DLNMAX is larger than the set value LVLMF, a drop in torque larger than combustion deterioration due to misfire is determined. Proceed to step 615 as it has occurred. In step 615, it is determined whether or not the maximum torque fluctuation amount DLNMAX is equal to or less than 2.5 times the set value LVLMF. Depending on the degree of misfire, depending on the degree, there is a case where a large drop in torque occurs that is not perceived by the driver without any combustion (hereinafter, referred to as a large misfire), or combustion is terminated on the way and cannot be perceived by the driver May cause a larger torque drop than the combustion deterioration (hereinafter referred to as a small misfire), and the maximum torque fluctuation amount D
If LNMAX is greater than 2.5 times the set value LVLMF, it is determined that a large misfire has occurred, and step 618 is performed.
, The misfire occurrence flag XMFIRE is set (XMFIRE ← 1), and the large misfire occurrence flag XMF
IREH is set (XMFIREH ← 1).

On the other hand, when the maximum torque fluctuation amount DLNMAX is larger than the set value LVLMF but is smaller than 2.5 times the set value LVLMF, in step 617, the misfire occurrence flag XMF
The IRE is set (XMFIRE ← 1), but the large misfire occurrence flag XMFIREH is reset (XMFIREH).
← 0). If the maximum torque fluctuation amount DLNMAX is equal to or smaller than the set value LVLMF, step 616 is executed.
, The misfire occurrence flag XMFIRE and the large misfire occurrence flag XMFIREH are both reset (XMF
IRE ← 0, XMFIREH ← 0).

Thus, the misfire level of the engine is determined, and then the routine proceeds to step 619, where the average torque fluctuation value DLNISM (= ｛DLNISM (1) + DLNISM), which is the average of the torque fluctuation values DLNISM (i) of each cylinder.
(2) + DLNISM (3) + DLNISM (4)｝ /
4) is calculated. That is, the average value of the torque fluctuation values DLNISM (i) of the respective cylinders when the torque fluctuation amount DLNSM (1) for the first cylinder is calculated 16 times and the torque fluctuation value DLNISM (1) of the first cylinder is calculated. Is calculated.
While the torque fluctuation amount DLN (1) of the first cylinder is accumulated 16 times, the torque fluctuation of the cylinder in which the calculation of the torque fluctuation amount is prohibited more than the number of times the calculation of the torque fluctuation amount of the first cylinder is prohibited. Since the value DLNISM (i) has not been newly calculated in step 609, the previous value is used for calculating the average torque fluctuation value DLNISM.

Next, at step 620, it is determined whether or not the misfire occurrence flag XMFIRE is set (XMFIRE = 1). If the misfire occurrence flag XMFIRE is not set, the routine proceeds to step 625, where the misfire occurrence interval count value CMFINT is set to 15 Above and the average torque fluctuation value DLNISM is equal to the target torque fluctuation value LVLLLF.
It is determined whether it is larger than B or not. The misfire occurrence interval count value CMFINT is a value that is incremented by 1 every time the following step 628 is passed. That is, as a result of the determination of the misfire level of the engine as described above, a misfire occurs in any cylinder. If not, it is incremented by one. The target torque fluctuation value LVLLFB is
It is calculated from the engine load Q / N and the engine speed N based on the map shown in FIG. The misfire occurrence interval count value CMFINT is 15 or more and the average torque fluctuation value DLN
When the ISM is larger than the target torque fluctuation value LVLLFB, the routine proceeds to step 626, where the misfire warning flag XMFI
REW is reset (XMFIREW ← 0), and in step 627, the lean correction execution count value CNOMF
Is incremented by one, and in step 628, the misfire occurrence interval count value CMFINT is incremented by one. On the other hand, the misfire occurrence interval count value CMFI
NT is less than 15 or average torque fluctuation value DLNIS
When M is equal to or smaller than the target torque fluctuation value LVLLFB, the routine jumps to step 627 without resetting the misfire warning flag XMFIREW. Misfire warning flag X
MFIREW and lean-side correction execution count value CNO
The MF is used in the FLLB calculation routine shown in FIGS. 32 to 34, and will be described in this routine. Next, at step 629, the torque fluctuation timing flag XHENDO is set (XHENDO ←
1) The processing cycle is completed. The torque variation timing flag XHENDO will be described later. Torque fluctuation amount DLN (i) used in this routine
Is very accurate calculated in consideration of the torsional vibration of the engine drive system and the torsional vibration of the crankshaft itself by the aforementioned torque fluctuation calculation routine. However, in order to simplify the control, the torque variation DLN
As (i), even if the torque variation calculated without considering one or both of these torsional vibrations is used,
It is possible to determine a certain level of misfire.

On the other hand, when the misfire occurrence flag XMFIRE is set at step 620, the routine proceeds to step 621, where the misfire occurrence interval count value CMFINT is set.
Is determined to be 3 or less, and if the misfire occurrence interval count value CMFINT is 3 or less, step 62 is executed.
2 to set the misfire warning flag XMFIREW (X
MFIREW ← 1), and proceeds to step 623. If the misfire occurrence interval count value CMFINT is greater than 3, the routine proceeds to step 623 without setting the misfire warning flag XMFIREW. In step 623, the lean-side correction execution count value CNOMF is reset (CNOMF ←
0), the misfire occurrence interval count value CMFINT is reset (CMFINT ← 0) in step 624, and then the torque fluctuation timing flag XHENDO is set (XHENDO ← 1) in step 629, and the processing cycle is completed.

Next, the FLLF shown in FIGS.
The B calculation routine will be described. First, at step 701, it is determined whether or not the condition for updating the lean limit feedback correction coefficient FLLFB is satisfied. For example, in the aforementioned torque fluctuation value calculation and misfire level determination routine, the torque fluctuation timing flag XH
When ENDO is not set (XHENDO =
0), or during the warm-up operation, or when the operating state of the engine is not in the learning region surrounded by the broken line in FIG. 5, it is determined that the update condition is not satisfied. Is determined to be. When the update condition is not satisfied, the processing cycle is completed, and when the update condition is satisfied, the process proceeds to step 702.

In step 702, a target torque fluctuation value LVLLFB is calculated from the engine load Q / N and the engine speed N based on the map shown in FIG. Next, at step 703 and step 704, the target torque fluctuation value L
Fluctuation discrimination value DH (n), DL according to VLLFB
Based on (n), the torque fluctuation level LV represented by the following equation:
LH (n) and LVLL (n) are calculated. LVLH (n) = LVLLFB + DH (n) LVLL (n) = LVLLFB + DL (n)

Here, the variation discrimination values DH (n) and D
L (n) is predetermined as shown in FIG. That is, as can be seen from FIG.
For (n), three positive values are defined, and DH
(3)> DH (2)> DH (1). Further, these DH (1), DH (2), DH (3) gradually increase as the target torque fluctuation value LVLLFB increases. On the other hand, three negative values are defined for DL (n), and DL (1)> DL (2)> DL (3)
Has the relationship Furthermore, these DL (1), DL
The absolute values of (2) and DL (3) are the target torque fluctuation values LVL
It gradually increases as the LFB increases.

Now, suppose that the target torque fluctuation value LVLLFB calculated in step 702 is a value indicated by a broken line. In this case, in step 703, the values obtained by adding DH (1), DH (2), and DH (3) on the broken line to the target torque fluctuation value LVLFB are the torque fluctuation levels LVLH (1), LVLH (2), and LVLH ( 3), and in step 704, DL (1), DL on the broken line
The values obtained by adding (2) and DL (3) to the target torque fluctuation value LVLLFB are the torque fluctuation levels LVLL (1) and LLL, respectively.
VLL (2) and LVLL (3).

On the other hand, as shown in FIG.
Area between the torque fluctuation levels LVLH (n) and LVLL (n)
Feedback correction value + a₁, + A_Two, + A
_Three, + A_Four, -B₁, -B_Two, -B_Three, -B_FourIs predetermined
For example, when the torque fluctuation level is LVLH
Feed for the area between (1) and LVLH (2)
The back correction value is + a_TwoBecomes These feedback supplements
Positive value is + a_Four> + A_Three> + A_Two> + A₁And -b
₁> -B_Two> -B_Three> -B_FourIt is. As shown in FIG.
Each feedback correction value + a₁, + A_Two, + A_Three, +
a_Four, -B₁, -B _Two, -B_Three, -B_FourIs shown in FIG.
In the corresponding area.

In steps 703 and 704, the torque fluctuation levels LVLH (n) and LVLL are respectively set.
When (n) is calculated, the routine proceeds to step 705, where the average engine speed N _AVE and the intake air amount obtained in step 409 of the routine for calculating the average engine speed and the average intake air amount shown in FIG. The lean limit feedback correction coefficient FLLFB _ij to be updated is determined based on the average value Q _AVE of the learning region shown in FIG.
Next, at step 707, the torque fluctuation timing flag XHENDO is reset (XHENDO ← 0), and the update condition at step 701 is satisfied until a new torque fluctuation value is calculated and a new misfire level is determined in the above-described routine. Not to be.

Next, at step 708, it is determined whether or not the misfire occurrence flag XMFIRE is set (XMFIRE = 1). When the misfire occurrence flag XMFIRE is reset (XMFIRE = 0), that is,
If it is determined in the aforementioned torque fluctuation value calculation and misfire level determination routine that no misfire has occurred, the routine proceeds to step 709, where the average torque fluctuation value D calculated in the aforementioned torque fluctuation value calculation and misfire level determination routine is obtained.
It is determined which of the torque fluctuation levels LVLH (n) and LVLL (n) shown in FIG. 27 (B) is LNISM. Next, at step 710, the corresponding feedback correction amount DLFB is calculated. For example, now, the target torque fluctuation value LVLLFB is a value indicated by a broken line in FIG. 27 (A), and the calculated average torque fluctuation value DLN
ISM is LVLH (1) and LVLH in FIG.
(2), that is, the target torque fluctuation value LVL
The deviation of the average torque fluctuation value DLNISM with respect to LFB is DH (1) and DH (2) on the broken line in FIG.
In the case where the feedback correction value DLFB is between
are a _2.

Next, at step 711, it is determined whether or not the average torque fluctuation value DLNISM is equal to or greater than the target torque fluctuation value LVLLFB, and the average torque fluctuation value DLNIFB is determined.
When SM is equal to or greater than the target torque fluctuation value LVLLFB, the routine proceeds to step 715, where the positive limit feedback correction value DLFB calculated in step 710 is added to the lean limit feedback correction coefficient FLLFB _ij . As a result, the air-fuel ratio becomes smaller, so that the torque fluctuation amount of each cylinder is reduced.

On the other hand, if the average torque fluctuation value DLNISM is smaller than the target torque fluctuation value LVLLFB in step 711, the flow advances to step 712 to determine whether or not the misfire warning flag XMFIREW has been reset in the aforementioned torque fluctuation value calculation and misfire level determination routine. Is determined. If the misfire warning flag XMFIREW is reset, that is, even if the average torque fluctuation value DLNISM becomes larger than the target torque fluctuation value LVLLFB, a misfire has not occurred for a sufficient period (for example, 240 cycles), and the average torque fluctuation has not occurred. When the value DLNISM is smaller than the target torque fluctuation value LLLFB, the routine proceeds to step 716, where the lean limit feedback correction coefficient FLLFFB is performed.
_The negative feedback correction value DLFB calculated in step 710 is added to ij. As a result, the air-fuel ratio increases, so that the amount of torque fluctuation in each cylinder is increased.

On the other hand, in step 712, a misfire warning
If the flag XMFIREW is set,
Go to step 713 and feed calculated in step 710
The back correction value DLFB is -b_TwoThe following and
The torque fluctuation value calculation and misfire level judgment routine
The side correction execution count value CNOMF is 8 or more
It is determined whether or not it is. The misfire warning flag XMFIREW
That is, a misfire has occurred
No, but in the past misfires were relatively short intervals (for example,
(48 cycles or less)
The back correction value DLFB is -b_TwoIs the current
The average torque fluctuation value at
Relatively small and lean-side correction execution count value CN
When OMF is 8 or more, that is, for a certain period (eg,
For example, 128 cycles) when misfire has not occurred
Proceeds to step 717 and proceeds to lean limit feedback.
Correction coefficient FLLFB_ijIs multiplied by (1-β),
In step 719, the lean-side correction execution count value C
NOMF is reset (CNOMF ← 0). That is,
In the past, misfires occurred continuously at relatively short intervals
No misfire occurred for 128 cycles, and the average
The torque fluctuation value DLNISM is compared with the target torque fluctuation value.
When relatively small, lean limit feedback compensation
Coefficient FLLFB _ijIs β% every 128 cycles (eg,
0.04%), and the air-fuel ratio gradually decreases
Is corrected.

On the other hand, in step 708, the misfire occurrence flag XMFIRE is set (XMFIRE =
In step 1), the routine proceeds to step 714, where the lean limit feedback correction coefficient FLLFB _ij is multiplied by (1 + α).
FIRE is reset (XMFIRE ← 0). That is, when a misfire has occurred, the lean limit feedback correction coefficient FLLFB _ij becomes α% every 16 cycles.
(For example, 0.8%), and the air-fuel ratio is largely corrected to the rich side.

Next, the routine proceeds to step 720, where the large misfire occurrence flag XMFIREH is set (XMFIREH = 1).
It is determined whether or not it has been performed. If no major misfire has occurred, the process jumps to step 725, but if a major misfire has occurred, the process proceeds to step 721 and proceeds to step 71.
4 as well as the lean limit feedback correction coefficient FL
LFB _ij is multiplied by (1 + α). Thus, when a large misfire occurs, the lean limit feedback correction coefficient FLLFB _ij is increased by α% (for example, 0.8%) every 16 cycles in addition to step 714, that is, as a result, About 1.6% increased,
The air-fuel ratio is corrected to a considerably large value on the rich side.

Next, the routine proceeds to step 722, where the lean limit feedback correction coefficient F when a large misfire occurs
LLFB _ij is the corresponding lower guard value FLLFBMIN
_It is determined whether it is greater than _ij . Lower guard value FLL
FBMIN _ij is set for each lean limit feedback correction coefficient FLLFB _ij . This lean limit feedback correction coefficient is a value FLLFB _k−2 before correction to the rich side in step 714,
When the lower guard value FLLFBMIN _{ij is} larger than
Proceeding to step 723, the lower guard value FLLFBM
IN _ij is a lean limit feedback correction coefficient FLLFB _ij when a large misfire occurs.
Proceed to 4. On the other hand, when the lean limit feedback correction coefficient FLLFB _ij at the time of the occurrence of the large misfire is equal to or smaller than the corresponding lower limit guard value FLLFBMIN _ij , the routine jumps to step 724. Next, at step 724, the large misfire occurrence flag XMFIREH is reset, and the routine proceeds to step 725.

In step 725, in this routine
The calculated lean limit feedback correction coefficient FL
LFB_ijIs the corresponding upper guard value FLLFFBMAX_ij
Is greater than or equal to the lean limit feed.
Back correction coefficient FLLFB_ijIs the corresponding upper guard value
FLLFBMAX_ijIf greater, step 726
To the lean limit feedback correction coefficient FLL
FB_ijIs the corresponding upper guard value FLLFFBMAX_ijWhen
And the processing routine is completed. On the other hand, to step 725
The lean limit fee calculated in this routine
Dbackback correction coefficient FLLFB_ijBut the corresponding upper limit guard
Value FLLFBMAX_ijIf not, step 7
Proceeding to 27, the lean limit feedback correction coefficient F
LLFB _ijIs the corresponding lower guard value FLLFBMIN
_ijIt is determined whether it is smaller than
Dbackback correction coefficient FLLFB_ijBut the corresponding lower guard
Value FLLFBMIN_ijIf it is more than the processing routine
Completed, but the lean limit feedback
Number FLLFFB_ijIs the corresponding lower guard value FLLFBM
IN_ijIf smaller, proceed to step 728 to reset
Limit feedback correction coefficient FLLFB_ijIs
Corresponding lower guard value FLLFBMIN_ijTo be processed
Complete the routine.

As described above, once a large misfire occurs, the lower limit guard value FLLFBMIN _ij is raised to the value of the lean limit feedback correction coefficient at the time of the large misfire, and thereafter the misfire does not occur. Even if the limit feedback correction coefficient FLLFB _ij is reduced, it is prevented from being lower than the lean limit feedback correction coefficient at the time of occurrence of a large misfire, that is, the air-fuel ratio is not made leaner than the air-fuel ratio at the time of occurrence of a large misfire. It is possible to prevent a large misfire from occurring. Since the likelihood of misfire greatly changes depending on weather conditions during engine operation, the lower guard value thus raised is maintained during engine operation and returned to the initial value after the engine stops. Has become. In the above description, the lower limit guard value is raised to the value of the lean limit feedback correction coefficient when a large misfire occurs. However, the lower limit guard value is raised to the lean limit feedback correction coefficient corrected in step 714 or step 721. Thus, it is possible to more reliably prevent the occurrence of a large misfire again.

When the calculation condition of the torque fluctuation value is not satisfied as shown in the routine shown in FIG. 24, at step 407, the DLNISM for all the cylinders is set.
(I) is LVLLFB, and thus the average torque fluctuation value DLNISM is also the target torque fluctuation value LVLLFB. Accordingly, at this time, the lean limit feedback correction coefficient FLLFB _ij is not updated.

Next, a routine for calculating the fuel injection time will be described with reference to FIG. Referring to FIG. 35, first, in step 801, the basic fuel injection time TP is calculated from the map shown in FIG. Then step 8
In 02, it is determined whether or not the operating state is such that the lean operation should be performed. When the vehicle is in the operating state in which the lean operation is to be performed, the routine proceeds to step 803, where the stoichiometric air-fuel ratio feedback correction coefficient F
The value of AF is fixed at 1.0. Then step 804
Then, the lean correction coefficient FLEAN is calculated from the map shown in FIG.
Is read. Next, at step 809, the fuel injection time TAU is calculated based on the following equation. TAU = TP ・ FLEAN ・ FLLFB ・ FAF + TA
UV

On the other hand, when it is determined in step 802 that the operating state is not such that lean operation should be performed, that is, when the air-fuel ratio should be set to the stoichiometric air-fuel ratio, step 806 is executed.
Then, the lean correction coefficient FLEAN is fixed at 1.0, and then at step 807, the lean limit feedback correction coefficient FLLFB is fixed at 1.0.
Next, at step 808, the stoichiometric air-fuel ratio feedback correction coefficient FAF is controlled based on the output signal of the air-fuel ratio sensor 17 so that the air-fuel ratio becomes the stoichiometric air-fuel ratio. Next, the routine proceeds to step 809, where the fuel injection time TAU is calculated.

FIG. 36 shows a lean limit feedback correction coefficient FL for a change in the torque fluctuation DLN in the air-fuel ratio control during the lean operation according to the above-described embodiment.
It is a time chart which shows a change of LFB. This will be described below. When the torque variation DLN becomes larger than 2.5 times the set value LVLMF (T1), it is determined that a large misfire has occurred, and the lean limit feedback correction coefficient F
LLFB is increased by about 1.6%. At this point, the misfire warning flag XMFIREW has been reset and FL
In step 715 or 717 of the LFB calculation routine, the lean limit feedback correction coefficient is updated by the feedback correction value determined based on the average torque fluctuation value DLNISM.

Next, the torque variation DLN is set to the set value LVL.
If it is smaller than 2.5 times the MF but larger than the set value LVLMF (T2), it is determined that a small misfire has occurred, and the lean limit feedback correction coefficient FLLFB is set to about 0.
Increased by 8%. When a small misfire occurs again in a relatively short period of about 48 cycles or less (T3), the lean limit feedback correction coefficient FLLFB is increased by about 0.8%, and the misfire warning flag XMFIREW is set. Thereafter, if the average torque fluctuation value DLNISM is equal to or greater than the target torque fluctuation value LVLLFB, the average torque fluctuation value DLNIS is determined in step 715 of the FLLFB calculation routine.
Although the lean limit feedback correction coefficient is increased by the feedback correction value determined based on M,
The average torque fluctuation value DLNISM is equal to the target torque fluctuation value LV.
If it is smaller than LLFB, the lean limit feedback correction coefficient FLLFB is reduced by 0.04% about every 128 cycles by the processing in step 716 of the FLLFB calculation routine.

Thereafter, if a small misfire occurs again (T
4), lean limit feedback correction coefficient FLLF
B is increased by about 0.8%, after which the average torque variation D
If LNISM is smaller than the target torque fluctuation value LLLFB, the lean limit feedback correction coefficient FLLF
B is reduced by 0.04% about every 128 cycles. Such a lean limit feedback correction coefficient FLLF
Even if the average torque fluctuation value DLNISM becomes larger than the target torque fluctuation value LVLLFB due to the decrease of B, approximately 240
When the misfire does not occur for more than one cycle (T5), the misfire warning flag XMFIREW is reset and FLLFB
In step 715 or 717 of the calculation routine, the lean limit feedback correction coefficient is updated by the feedback correction value determined based on the average torque fluctuation value DLNISM. When the lean limit feedback correction coefficient FLLFFB is updated and the aforementioned 0.04% decrease, the lean limit feedback correction coefficient F
LLFB is guarded at the lower limit by the feedback correction coefficient FLLFBMIN at the time of the occurrence of the large misfire (T1).

In contrast to the air-fuel ratio control, the ignition timing control is general. The optimum ignition timing _IGmn is read from the map shown in FIG. 38 based on the engine load Q / N and the engine speed N. It is implemented. The optimal ignition timing is
Generally referred to as MBT, it is the ignition timing on the most retarded side that generates the maximum torque. In the map shown in FIG. 38, this ignition timing is determined for each engine operating state determined by the engine load Q / N and the engine speed N. It is set in consideration of the air-fuel ratio at the time. Thus, the misfire can be reliably suppressed by the air-fuel ratio control according to the misfire level, but the misfire can also be suppressed by controlling the ignition timing. Such an ignition timing control will be described below.

FIG. 37 shows an ignition timing calculation routine for this purpose. In this routine, first, step 9
01, the engine load Q / N from the map shown in FIG.
And engine speed on the basis of the N optimum ignition timing IGmn are read, then in step 902, the ignition timing correction coefficient K _ij based on the intake air amount Q and engine speed N from the map shown in FIG. 39 is read, then steps 90
In 3, the ignition timing correction coefficient K is set to the optimum ignition timing IG _mn.
ij is multiplied to obtain the actual ignition timing IG.

In the map of the ignition timing correction coefficient K _ij shown in FIG. 39, the ignition timing correction coefficient K _ij is set for each learning region similar to the map of the lean limit feedback correction coefficient FLLFB shown in FIG. Things. Also, the ignition timing correction coefficient _K00 is set in a region outside the entire learning region, that is, in a region where the operation at the stoichiometric air-fuel ratio is performed. Initially, all the ignition timing correction coefficients K are set to 1. For ignition timing correction coefficient K ₀₀ of the area operated at the stoichiometric air-fuel ratio is performed is maintained to 1 without being updated, in the case of the operation at the stoichiometric air-fuel ratio, the ignition timing calculating routine shown in FIG. 37 , The optimum ignition timing _IGmn is used as the actual ignition timing without correction.

On the other hand, each ignition in the region where the lean operation is performed
Timing correction coefficient K_ijIs the FLLF shown in FIGS.
Be updated from 1 by the calculation routine of B and K
The ignition timing calculation routine shown in FIG.
At the optimal ignition timing IG _mnIs the advance side (ignition timing supplement
Positive coefficient K_ijIs updated to be greater than 1) or the retard side
(Ignition timing correction coefficient K_ijIs updated to less than 1
Is corrected. FLLFFB shown in FIGS. 40 to 42,
The calculation routine of K is based on the FLL described above with reference to FIGS.
This is performed instead of the FB calculation routine.
The corresponding steps are indicated by numbers with dashes.
ing. Only the differences between the two routines are described below
I do.

In the routine for calculating FLFLFB, K in step 713 ', the misfire warning flag XMFIRE is set.
W is set, that is, a misfire has not occurred at present, but a misfire has occurred continuously at relatively short intervals (for example, 48 cycles or less) in the past, and the feedback correction value DLFB is -b. ₂ or less, that is, when the current average torque fluctuation value is relatively smaller than the target torque fluctuation value and the lean-side correction execution count value CNOMF is 8 or more, that is, for a certain period (for example, when the cycle) misfire has not occurred, step 717 'in the ignition timing correction coefficient K _ij proceeds to the multiplied (1 + gamma), the step 719' lean correction execution count CNOMF in is reset (CNOMF ← 0) You. That is, even if misfires occur continuously at relatively short intervals in the past, no misfires have occurred for 128 cycles, and the average torque fluctuation value DLNI
When SM is relatively small with respect to the target torque fluctuation value,
The ignition timing correction coefficient K _ij is increased by γ% every 128 cycles, and the optimum ignition timing IGmn is corrected to the advanced side.

On the other hand, in step 708 ', the misfire occurrence flag XMFIRE is set (XMFIRE).
= 1), the routine proceeds to step 714 ′, where the ignition timing correction coefficient K _ij is multiplied by (1−δ).
At 8 ', the misfire occurrence flag XMFIRE is reset (XMFIRE ← 0). That is, when a misfire occurs, the ignition timing correction coefficient K _ij becomes δ every 16 cycles.
%, And the optimum ignition timing IGmn is corrected to the retard side.

Next, the routine proceeds to step 720 ', in which the large misfire occurrence flag XMFIREH is set (XMFIREH =
1) It is determined whether or not it has been performed. When a large misfire has not occurred, the process jumps to step 725 '. However, when a large misfire has occurred, the process proceeds to step 721', and similarly to step 714 ', the ignition timing correction coefficient _Kij contains (1--
δ) is multiplied. As described above, when a large misfire occurs, the ignition timing correction coefficient K _ij is further reduced by δ% every 16 cycles in addition to step 714 ′, that is, by about 2δ% as a result, and the optimal Ignition timing IG
mn is largely corrected to the retard side.

Next, the routine proceeds to step 722 ', where a great misfire occurs.
Ignition timing correction coefficient K when it occurs _ijBut the corresponding upper limit
Guard value KMAX_ijIt is determined whether or not it is smaller. Up
Limit guard value KMAX_ijIs the ignition timing correction coefficient K_ijSet for each
Is defined. This ignition timing correction coefficient is calculated in step 71
4 ', the value K before correction to the retard side_k-2In
The upper limit guard value KMAX_ijWhen smaller,
Proceeding to step 723, this upper limit guard value KMAX_ijIs large
Ignition timing correction coefficient K when misfire occurs_ijAnd
Proceed to step 724 '. On the other hand, when a major misfire
Ignition timing correction coefficient K_ijIs the corresponding upper guard value KMA
X_ijIf so, jump to step 724 '.
You. Next, at step 724 ', the large misfire occurrence flag XM
FIREH is reset and goes to step 725 '.

In steps 725 'to 728',
After the upper and lower limit guard processing of the lean limit feedback correction coefficient FLLFB is performed, step 72 is executed.
In 9 ', the ignition timing correction coefficient K _ij is whether greater than the corresponding upper limit guard value KMAX _ij is determined, the ignition timing correction coefficient K _ij is the corresponding upper limit guard value KMAX _ij
Ignition timing correction coefficient K _ij proceeds to step 730 'when larger completes is a corresponding upper guard value KMAX _ij routine. On the other hand, when the ignition timing correction coefficient K _ij is equal to or smaller than the corresponding upper guard value KMAX _ij in step 729 ′, the routine proceeds to step 731 ′, where the ignition timing correction coefficient K _ij is smaller than the corresponding lower guard value KMIN _ij. If the ignition timing correction coefficient K _ij is equal to or greater than the corresponding lower limit guard value KMIN _ij , the processing routine is completed, but the ignition timing correction coefficient K _ij is set to a value smaller than the corresponding lower limit guard value KMIN _ij . when small, the ignition timing correction coefficient K _ij proceeds to step 732 'completes the set to the corresponding lower limit guard value KMIN _ij routine.

[0140] Thus, once a large misfire occurs, the upper limit guard value KMAX _ij of the ignition timing correction coefficient K is pulled up to the value of the ignition timing correction coefficient at the time of large-misfiring,
After that, the misfire no longer occurs and step 71
Even if the ignition timing correction coefficient K _ij is increased in 6 ′, it is prevented from exceeding the ignition timing correction coefficient at the time of the occurrence of a large misfire, that is, the optimal ignition timing is advanced to the ignition timing at the time of the occurrence of the large misfire. It is possible to prevent the occurrence of a large misfire again. Since the likelihood of misfire greatly changes depending on weather conditions during operation of the engine, the lowered upper guard value is maintained during operation of the engine and returned to the initial value after the engine stops. Has become. In the above description, the upper limit guard value is reduced to the value of the ignition timing correction coefficient when a large misfire occurs, but the upper limit guard value is reduced to the ignition timing correction coefficient corrected in step 714 ′ or 721 ′. Thus, it is possible to more reliably prevent the occurrence of a large misfire again.

When the calculation condition of the torque fluctuation value is not satisfied, as shown in the routine shown in FIG. 24, at step 407, the DLNISM for all the cylinders is set.
(I) is LVLLFB, and thus the average torque fluctuation value DLNISM is also the target torque fluctuation value LVLLFB. Accordingly, at this time, the ignition timing correction coefficient K _ij is not updated.

As described above, the lean limit feedback correction coefficient FLLFB is updated in accordance with the torque fluctuation value DLNISM.
In the event of a misfire, instead of updating the lean limit feedback correction coefficient FLLFB to correct the air-fuel ratio to the rich side, the ignition timing is corrected to the retard side based on the same concept as described above. Thus, it is possible to reliably suppress misfire without deteriorating the fuel consumption rate due to the enrichment of the air-fuel ratio.

[0143]

According to the air-fuel ratio control apparatus for an internal combustion engine according to the present invention, the misfire level is determined in accordance with the magnitude of the driving force fluctuation generated by at least one cylinder, and the misfire can be reliably detected. In addition, since the air-fuel ratio or the ignition timing is controlled according to the misfire level, misfire can be reliably suppressed.

[Brief description of the drawings]

FIG. 1 is an overall view of an internal combustion engine.

FIG. 2 is a diagram showing a map of a basic fuel injection time.

3 is a diagram showing a generation amount and the torque fluctuation of the NO _x.

FIG. 4 is a diagram showing a map of a lean correction coefficient.

FIG. 5 is a diagram showing a map of a lean limit feedback correction coefficient.

FIG. 6 shows elapsed times Ta (i) and T of a 30 ° crank angle.
It is a time chart which shows the change of b (i).

FIG. 7 is a time chart showing a change in elapsed time Ta (i) at a 30 ° crank angle.

FIG. 8 shows elapsed times Ta (i) and T of a 30 ° crank angle.
It is a time chart which shows the change of b (i).

FIG. 9 shows elapsed times Ta (i) and T of a 30 ° crank angle.
It is a time chart which shows the change of b (i).

FIG. 10 is a time chart showing changes in elapsed time Ta (i) at a 30 ° crank angle.

FIG. 11 is a diagram showing a relationship between elapsed time differences DTa (i) and Kb (i).

FIG. 12 is a time chart showing changes in elapsed time Ta (i) during deceleration operation.

FIG. 13 is a time chart showing a change of an elapsed time difference DTa (i).

FIG. 14 is a flowchart showing an interrupt routine.

FIG. 15 shows an elapsed time difference DTa (i) and an elapsed time Tb.
It is a flowchart for calculating (i).

FIG. 16 shows an elapsed time difference DTa (i) and an elapsed time Tb.
It is a flowchart for calculating (i).

FIG. 17 is a flowchart for calculating KTa (i).

FIG. 18 is a flowchart for calculating KTb (i).

FIG. 19 is a flowchart for checking permission for calculating a torque fluctuation amount.

FIG. 20 is a flowchart for checking permission for calculating a torque fluctuation amount.

FIG. 21 is a flowchart for checking permission of calculation of a torque fluctuation amount.

FIG. 22 shows changes in elapsed time difference DTa (i) and flag XM.
It is a time chart which shows a change of XREC and XMREC.

FIG. 23 is a flowchart for calculating a torque fluctuation amount.

FIG. 24 is a flowchart for calculating an average engine speed and an average intake air amount.

FIG. 25 is a diagram showing calculation timings of various values.

FIG. 26 is a diagram showing a target torque fluctuation value.

FIG. 27 is a diagram showing fluctuation amount determination values DH (n) and DL (n) and torque fluctuation levels LVLH (n) and LVLL (n).

FIG. 28 is a flowchart showing a main routine.

FIG. 29 is a flowchart for calculating a torque fluctuation value and determining a misfire level.

FIG. 30 is a flowchart for calculating a torque fluctuation value and determining a misfire level.

FIG. 31 is a flowchart for calculating a torque fluctuation value and determining a misfire level.

FIG. 32 is a flowchart for calculating a lean limit feedback correction coefficient.

FIG. 33 is a flowchart for calculating a lean limit feedback correction coefficient.

FIG. 34 is a flowchart for calculating a lean limit feedback correction coefficient.

FIG. 35 is a flowchart for calculating a fuel injection time.

FIG. 36 is a time chart showing changes in the lean limit feedback correction coefficient with respect to changes in the amount of torque fluctuation in air-fuel ratio control during lean operation.

FIG. 37 is a flowchart for calculating an ignition timing.

FIG. 38 is a view showing a map of an optimum ignition timing. is there.

FIG. 39 is a diagram showing a map of an ignition timing correction coefficient.

FIG. 40 is a flowchart for calculating a lean limit feedback correction coefficient and an ignition timing correction coefficient.

FIG. 41 is a flowchart for calculating a lean limit feedback correction coefficient and an ignition timing correction coefficient.

FIG. 42 is a flowchart for calculating a lean limit feedback correction coefficient and an ignition timing correction coefficient.

[Explanation of symbols]

 3. Surge tank 4. Fuel injection valve 8. Throttle valve 14. Rotor 15. Crank angle sensor

──────────────────────────────────────────────────続 き Continuation of the front page (51) Int.Cl. ⁷ Identification symbol FI F02P 5/15 F02P 5/15 B (56) References JP-A-60-27748 (JP, A) JP-A-60-27761 ( JP, A) JP-A-4-301161 (JP, A) JP-A-6-117291 (JP, A) JP-A-5-222989 (JP, A) JP-A-6-307285 (JP, A) JP Sho 63-302163 (JP, A) JP-A 3-51178 (JP, U) JP-B 7-33809 (JP, B2) (58) Fields investigated (Int. Cl. ⁷ , DB name) F02D 41 / 00-41/40 F02D 43/00 F02D 45/00 F02P 5/15

Claims (3)

(57) [Claims] 1. A control device for an internal combustion engine for controlling an air-fuel ratio to be leaner than a stoichiometric air-fuel ratio at least in a predetermined operation state, wherein a first crank angle range is set within a crank angle range from the end of a compression stroke to the beginning of an explosion stroke. And setting a second crank angle range within a middle crank angle range of an explosion stroke separated by a certain crank angle from the first crank angle range, and setting a second crank angle range of the crankshaft within the first crank angle range. 1 and the second angular speed of the crankshaft within the second crank angle range is detected, and the driving force variation generated by each cylinder based on the first angular speed and the second angular speed is calculated. A driving force variation calculating means for calculating, and an average value of the driving force variation of each cylinder calculated by the driving force variation calculating means is a target value. First air-fuel ratio control means for controlling the air-fuel ratio so as to obtain the driving force fluctuation amount; A misfire level judging means for judging the air-fuel ratio control by the first air-fuel ratio control means when a misfire occurs and the misfire level judged by the misfire level judging means exceeds a first level. A control device for an internal combustion engine, comprising: second air-fuel ratio control means for controlling an air-fuel ratio according to a misfire level. 2. When the misfire level determined by the misfire level determination means exceeds a second level which is larger than the first level, thereafter, the first misfire level is controlled by the first air-fuel ratio control means or the second air-fuel ratio control means. 2. The control device for an internal combustion engine according to claim 1, further comprising a guard means for guarding the air-fuel ratio from being leaner than the air-fuel ratio when the misfire level exceeds the second level in the air-fuel ratio control. 3. A control device for an internal combustion engine that controls an air-fuel ratio to be leaner than a stoichiometric air-fuel ratio at least in a predetermined operating state, wherein the first crank angle range is within a crank angle range from the end of a compression stroke to the beginning of an explosion stroke. And setting a second crank angle range within a middle crank angle range of an explosion stroke separated by a certain crank angle from the first crank angle range, and setting a second crank angle range of the crankshaft within the first crank angle range. 1 and the second angular speed of the crankshaft within the second crank angle range is detected, and the driving force variation generated by each cylinder based on the first angular speed and the second angular speed is calculated. A driving force variation calculating means for calculating, and an average value of the driving force variation of each cylinder calculated by the driving force variation calculating means is a target value. Air-fuel ratio control means for controlling the air-fuel ratio so as to obtain the driving force fluctuation amount; ignition timing control means for controlling the ignition timing so as to obtain the optimum ignition timing calculated based on the engine operating state; A misfire level judging means for judging a misfire level according to the magnitude of the driving force fluctuation amount of at least one cylinder calculated by the arithmetic means, and the optimum misfire level according to the misfire level determined by the misfire level judgment means. A control device for an internal combustion engine, comprising: ignition timing correction means for correcting an ignition timing.