GB2509351B

GB2509351B - Air-fuel ratio control system

Info

Publication number: GB2509351B
Application number: GB1316952.9A
Authority: GB
Inventors: Fujita Shinji; Takano Yuki; Nakano Shinya; Sakiyama Satoko
Original assignee: Honda Motor Co Ltd
Current assignee: Honda Motor Co Ltd
Priority date: 2012-09-28
Filing date: 2013-09-24
Publication date: 2019-09-18
Anticipated expiration: 2033-09-24
Also published as: JP2014070600A; ITTO20130725A1; GB201316952D0; GB2509351A; JP5981827B2

Description

AIR-FUEL RATIO CONTROL SYSTEM

The present invention relates to an air-fuel ratio control system, and inparticular to an air-fuel ratio control system suitable for use in a vehicle (such as amotorcycle) with an internal combustion engine.

In an automobile system for decontaminating exhaust gas from an internalcombustion engine (hereinafter "engine") by means of a catalytic device andexhausting it, it is desirable in view of environmental protection that the air-fuel ratioof exhaust gas of the engine is controlled to be an air-fuel ratio at which the exhaustgas decontamination capacity of the catalytic device is satisfactory. A system that performs such air-fuel ratio control is disclosed in JP3373724. In this document, the air-fuel ratio control system is configured so that acorrection factor is superimposed on fuel injection quantity, so as to remove the lagof the injection quantity acquired based upon a fuel injection volume map (havingengine speed, a throttle opening, vacuum and others as parameters) fordetermining the fuel injection quantity in the engine from target air-fuel ratio.

Specifically, an LAF sensor (a sensor that outputs a signal at a levelproportional to the oxygen content of exhaust gas in a large range of the oxygencontent (the air-fuel ratio) of exhaust gas) is installed upstream of a catalyticdecontaminating device arranged in an exhaust pipe of the engine, and an oxygensensor (an air-fuel ratio sensor) is installed downstream of the catalytic device. Apredicted value of air-fuel ratio after catalysis is calculated using a detection valueby the LAF sensor, and the correction factor is acquired using the predicted valueby a sliding mode controller, for example.

As LAF sensors are expensive, there is a desire to do away with the LAFsensor installed upstream of the catalytic device, because of the cost reduction ofthe system and the limit of space for arrangement in a motorcycle and othervehicles.

However, as an output value (SVO2) of the oxygen sensor (to be a desiredvalue of emission) converges on the desired value based upon the output value(SV02) input to the sliding mode controller (SMC) which models the intake/exhaustof the engine, the air-fuel ratio before catalysis cannot be measured when no LAFsensor is installed upstream of the catalytic device; therefore, the prediction of thetolerance and the age deterioration of the engine and the injection error of a fuelinjection valve in the modelled engine cannot be monitored, a predictable range of a predicted value of the output value (SV02) is enlarged, and much time may berequired for the convergence on the desired value by the sliding mode controller(SMC).

Besides, as gain in convergence by the sliding mode controller (SMC) alsohas a limit in adjustment, it is also conceivable that a predictive error of a predictedvalue of an output value (SV02) cannot be eliminated and the output value (SV02)cannot be converged on the desired value.

Depending upon the specifications of an internal combustion engine, asystem may be installed in the exhaust pipe for injecting secondary air, upstream ofa catalyst, to reduce the emission of exhaust gas. As the air-fuel ratio before entryinto the catalyst varies when secondary air is injected, the sliding mode controllermay have an effect upon the whole control system. Besides, generally, secondaryair is set to supply sufficient air so as to enable secure combustion of residual fuelin the exhaust pipe unless special feedback is made. As a result, when secondaryair is injected, the air-fuel ratio upstream of the catalyst is apt to be lean (oxygen isexcessive). Therefore, when secondary air is injected, the quantity of oxygenstored in the catalyst readily increases immediately after the injection of secondaryair and as a result, NOX decontamination performance may be deteriorated.

Moreover, as air also flows into the catalyst when fuel injection is cut while athrottle valve is fully closed except the injection of secondary air, the quantity ofoxygen stored in the catalyst increases and the same problem as that in theinjection of secondary air occurs. Similarly, when deceleration leaning is madewithout cutting fuel from a viewpoint of protecting the catalyst, the quantity ofoxygen stored in the catalyst also increases.

The present invention is made in view of such problems, and it is an objectof at least the preferred embodiments of the present invention to provide an air-fuelratio control system where air-fuel ratio can be optimized without installing an LAFsensor upstream of a catalytic device, the cost of the system can be reduced, theapplication of air-fuel ratio control to a motorcycle and others can be accelerated,and in addition, in the case of the injection of secondary air, fuel injection cut ordeceleration leaning, air-fuel ratio control can be also promptly restored afterrestoration to a normal state from a control state of these, and emissionperformance can be kept high.

According to a first aspect of the present invention, there is provided an air-fuel ratio control system comprising: a basic fuel injection map that specifies fuel injection quantity to an engine based upon at least parameters of engine speed andthrottle opening; air-fuel ratio sensing means which is provided downstream of acatalyst installed in an exhaust pipe of the engine and which senses air-fuel ratio;air-fuel ratio predicting means that predicts air-fuel ratio downstream of the catalyst;and correction factor calculating means that determines a correction factor for thefuel injection quantity based upon predicted air-fuel ratio from the air-fuel ratiopredicting means by feedback control, wherein: the air-fuel ratio predicting meanscalculates the predicted air-fuel ratio based upon at least real air-fuel ratio from theair-fuel ratio sensing means and a history of the correction factor; the air-fuel ratiocontrol system has adaptive model correcting means that superimposes a secondcorrection factor on the correction factor so as to make deviation between the realair-fuel ratio and the predicted air-fuel ratio which corresponds to the real air-fuelratio and which is predicted in the past as a predictive error zero; a secondary airpulse induction system that injects secondary air into an exhaust path upstream ofthe catalyst is provided; the feedback control is interrupted at a stage at which theinjection of secondary air is started; immediately after the injection of secondary airis finished, air-fuel ratio control is made to shift to rich control in an open loop; andafter the real air-fuel ratio transfers to a rich side, the feedback control is resumed.

With this arrangement, the quantity of oxygen stored in the catalyst as aresult of the injection of secondary air can be reduced at an early stage byinterrupting the feedback control after the injection of secondary air in the system inwhich the secondary air pulse induction system is installed, and performing richinjection control at the stage at which the injection of secondary air is finished. As aresult, the interrupted feedback control can be restored at an early stage andemission performance can be kept high.

According to a second aspect of the present invention, there is provided anair-fuel ratio control system comprising: a basic fuel injection map that specifies fuelinjection quantity to an engine based upon at least parameters of engine speed andthrottle opening; air-fuel ratio sensing means which is provided downstream of acatalyst installed in an exhaust pipe of the engine and which senses air-fuel ratio;air-fuel ratio predicting means that predicts air-fuel ratio downstream of the catalyst;and correction factor calculating means that determines a correction factor for thefuel injection quantity based upon predicted air-fuel ratio from the air-fuel ratiopredicting means by feedback control, wherein: the air-fuel ratio predicting meanscalculates the predicted air-fuel ratio based upon at least real air-fuel ratio from the air-fuel ratio sensing means and a history of the correction factor; the air-fuel ratiocontrol system has adaptive model correcting means that superimposes a secondcorrection factor on the correction factor so as to make deviation between the realair-fuel ratio and the predicted air-fuel ratio which corresponds to the real air-fuelratio and which is predicted in the past as a predictive error zero; the air-fuel ratiocontrol system has fuel cut control means that performs fuel injection halt controlwhile the throttle opening is closed; when the fuel injection halt control by the fuelcut control means is started, the feedback control is interrupted; immediately afterthe fuel injection halt control by the fuel cut control means is finished, air-fuel ratiocontrol is made to shift to rich control in an open loop; and after the real air-fuel ratiotransfers to a rich side, the feedback control is resumed.

As air containing oxygen flows into the catalyst when fuel injection haltcontrol is performed by the fuel cut control means by closing the throttle opening,the quantity of oxygen stored in the catalyst increases. However, with thisarrangement, as the feedback control is interrupted when the fuel injection haltcontrol by the fuel cut control means is started and rich injection control isperformed at the stage at which the fuel injection halt control is finished, thequantity of oxygen stored in the catalyst by the intake of air containing oxygenaccording to the fuel injection halt control can be reduced at an early stage. As aresult, the interrupted feedback control can be restored at an early stage andemission performance can be kept high.

According to a third aspect of the present invention, there is provided an air-fuel ratio control system comprising: a basic fuel injection map that specifies fuelinjection quantity to an engine based upon at least parameters of engine speed andthrottle opening; air-fuel ratio sensing means which is provided downstream of acatalyst installed in an exhaust pipe of the engine and which senses air-fuel ratio;air-fuel ratio predicting means that predicts air-fuel ratio downstream of the catalyst;and correction factor calculating means that determines a correction factor for thefuel injection quantity based upon predicted air-fuel ratio from the air-fuel ratiopredicting means by feedback control, wherein: the air-fuel ratio predicting meanscalculates the predicted air-fuel ratio based upon at least real air-fuel ratio from theair-fuel ratio sensing means and a history of the correction factor; the air-fuel ratiocontrol system has adaptive model correcting means that superimposes a secondcorrection factor on the correction factor so as to make deviation between the realair-fuel ratio and the predicted air-fuel ratio which corresponds to the real air-fuel ratio and which is predicted in the past as a predictive error zero; the air-fuel ratiocontrol system has deceleration leaning control means that performs decelerationleaning; when the deceleration leaning by the deceleration leaning control means isstarted, the feedback control is interrupted; immediately after the decelerationleaning by the deceleration leaning control means is finished, air-fuel ratio control ismade to shift to rich control in an open loop; and after the real air-fuel ratio transfersto a rich side, the feedback control is resumed.

Air containing oxygen flows into the catalyst as in the case of the fuel cutcontrol when the deceleration leaning by the deceleration leaning control means isstarted, and so the quantity of oxygen stored in the catalyst increases. However, j with this arrangement, as the feedback control is interrupted when the decelerationleaning by the deceleration leaning control means is started and rich injectioncontrol is performed at the stage at which the deceleration leaning is finished, thequantity of oxygen stored in the catalyst by the intake of air containing oxygenaccording to the deceleration leaning can be reduced at an early stage. As a result,the interrupted feedback control can be restored at an early stage and emissionperformance can be kept high.

Preferably, the feedback control is sliding mode control; and when thepredictive error exceeds a preset threshold prior to resumption of the feedbackcontrol, PID control is performed so as to make an error between the real air-fuelratio and a preset desired value zero.

Thus, when it is predicted that time is required for the convergence of slidingmode control by the injection of secondary air or fuel cut control or the intake of aircontaining oxygen according to the deceleration leaning, the convergence of thefeedback control can be accelerated by first performing PID control at a stage atwhich the injection of secondary air is finished or at a stage at which the fuel cutcontrol or the deceleration leaning is finished, and emission performance can bekept high.

In a further preferred form, the feedback control is resumed at a stage atwhich the predictive error is equal to or smaller than the preset threshold.

As the convergence of sliding mode control is secured when the predictiveerror is equal to or smaller than the threshold, emission performance can be kepthigh by resuming the feedback control.

Preferably, the air-fuel ratio control system further comprises a controllerthat controls at least the correction factor calculating means and the adaptive model correcting means, wherein: the adaptive model correcting means is provided withpredictive precision determining means that determines predictive precision basedupon the predictive error; and when deterioration of the predictive precision isdetermined in the predictive precision determining means in the resumption of thefeedback control, the PID control is performed without using the air-fuel ratiopredicting means so as to make the error between the real air-fuel ratio and thepreset desired value zero.

With this arrangement, PID control is performed so that the error betweenthe real air-fuel ratio and the preset desired value is made zero without using theair-fuel ratio predicting means when the deterioration of predictive precision isdetermined in the resumption of the feedback control. Thus, time until thepredictive precision is secured can be more reduced, compared with a case inwhich the air-fuel ratio predicting means is used and emission performance can bekept high.

Preferably, the air-fuel ratio control system further comprises a dedicatedPID controller, wherein: the feedback control is the sliding mode control; and whenthe error between the real air-fuel ratio and the preset desired value exceeds thepreset threshold prior to the resumption of the feedback control, the PID control isperformed in the PID controller so as to make the error zero.

With this arrangement, PID control is performed using the dedicated PIDcontroller so that the difference (the error) between the real air-fuel ratio and thedesired value is made zero prior to the resumption of the feedback control, and theerror can be converged on the threshold or less at a more early stage, comparedwith a case in which the predictive error in a main body of an air-fuel ratio controlleris used, and normal sliding mode control (control in a first sliding mode controller)can be resumed at a more early stage. Thus, emission performance can be kepthigh.

Preferred embodiments of the invention will now be described by way ofexample only and with reference to the accompanying drawings, in which:

Fig. 1 is a perspective view showing an example of a motorcycle in which anair-fuel ratio control system according to the invention is installed;

Fig. 2 is a block diagram showing an example of a control system of anengine of the motorcycle;

Fig. 3 is a functional block diagram showing a configuration of an air-fuelratio controller provided with a first air intake corresponding part;

Fig. 4 is a control block diagram showing the configuration of the air-fuelratio control system (the air-fuel ratio controller) in a first embodiment;

Fig. 5 is a control block diagram showing a configuration of a main body ofan air-fuel ratio controller in a comparative (prior art) example;

Fig. 6 is an explanatory drawing showing a predictive model by a predictor;

Fig. 7 is an explanatory drawing showing an operational concept of slidingmode control;

Fig. 8 is a block diagram showing a configuration of an adaptive modelcorrector;

Fig. 9 is a block diagram showing the specific configuration of the adaptivemodel corrector;

Fig. 10A is a characteristic diagram showing a variation of output of anoxygen sensor for air-fuel ratio A/F and Fig. 10B is a characteristic diagramshowing a variation of a first weighting component for real air-fuel ratio;

Fig. 11A is a characteristic diagram showing a variation of basic fuelinjection quantity for a throttle opening and Fig. 11B is a characteristic diagramshowing a variation of a second weighting component for the throttle opening;

Fig. 12A is a characteristic diagram showing a weighting function for enginespeed NE and Fig. 12B is a characteristic diagram showing a weighting function fora throttle opening TH;

Fig. 13 is an explanatory drawing for explaining a principle for acquiring acorrection factor based upon predictive error corrected quantity;

Fig. 14 is a control block diagram showing a configuration of a main body ofan air-fuel ratio controller in a first alternative embodiment;

Fig. 15 is a control block diagram showing a configuration of a main body ofan air-fuel ratio controller in a second alternative embodiment;

Fig. 16 is a control block diagram showing a configuration of a main body ofan air-fuel ratio controller in a third alternative embodiment;

Fig. 17 is a control block diagram showing a configuration of a main body ofan air-fuel ratio controller in a fourth alternative embodiment;

Fig. 18 is a control block diagram showing a configuration of a main body ofan air-fuel ratio controller in a fifth alternative embodiment;

Fig. 19 is a flowchart showing the processing operation of a first air intakecorresponding part;

Fig. 20 is a timing chart showing an execution period of the injection ofsecondary air or FC control or deceleration leaning, a rich spike control period (aresidual oxygen processing period of a catalytic device), the variation of fuelinjection quantity per unit time in rich spike control, the variation of the quantity ofoxygen stored in the catalytic device, and the variation of an output value (SVO2)from the oxygen sensor;

Fig. 21 is a functional block diagram showing a configuration of an air-fuelratio controller provided with a second air intake corresponding part; and

Fig. 22 is a flowchart showing the processing operation of the second airintake corresponding part.

Referring to Figs. 1 to 22, an embodiment in which an air-fuel ratio controlsystem according to the present invention is applied to a motorcycle as an examplevehicle will be described below.

First, the motorcycle 12 in which the air-fuel ratio control system 10 in thisembodiment is mounted will be described, referring to Fig. 1.

The motorcycle 12 is configured by coupling the front 14 of a vehicle bodyand the rear 16 of the vehicle body via a low floor panel 18 as shown in Fig. 1. Ahandlebar 20 is turnably attached to the upper part of the front 14 of the vehiclebody and a front wheel 22 is journaled to the lower part. A seat 24 is attached tothe upper part of the rear 16 of the vehicle body and a rear wheel 26 is journaled tothe lower part.

An intake pipe 30 and an exhaust pipe 32 are provided on an engine 28 ofthe motorcycle 12, as schematically shown in Fig. 2, and the intake pipe 30 isarranged between the engine 28 and an air cleaner 34. A throttle valve 38 isprovided in a throttle body 36 provided to the intake pipe 30. A fuel injection valve40 is provided between the engine 28 and the throttle body 36 on the intake pipe30.

The throttle valve 38 is turned according to the operation of a throttle grip 42(see Fig. 1), and its turned quantity (an angle of the throttle valve 38) is detected bya throttle sensor 44. Air volume supplied to the engine 28 is varied by opening orclosing the throttle valve 38 according to the operation of the throttle grip 42 by arider. A coolant temperature sensor 46 that detects the temperature of enginecooling water is provided on the engine 28, and a PB sensor 48 that detects intakeair pressure (intake vacuum) is provided on the intake pipe 30. There are also provided a secondary air pulse induction system 1000 which is provided upstreamof a catalytic device 50 installed in the exhaust pipe of the engine 28, and whichtakes air from the air cleaner 34 into the exhaust pipe as secondary air, and anoxygen sensor 52 (air-fuel ratio detection means) which is provided downstream ofthe catalytic device 50 installed in the exhaust pipe of the engine 28 and whichdetects air-fuel ratio downstream of the catalytic device 50. An oxygen contentdetected by the oxygen sensor 52 is equivalent to the real air-fuel ratio of exhaustgas after the exhaust gas passes through the catalytic device 50.

Furthermore, a vehicle speed sensor 56 that detects vehicle speed basedupon the revolution speed of an output gear of reduction gears 54 is provided onthe engine 28. A starter switch 58 is a switch for starting the engine 28 by theoperation of an ignition key. Further, an atmospheric pressure sensor 60 isprovided in a position remote from the intake pipe 30 of the air cleaner 34.

An engine control unit (ECU 62) is provided with an Al (secondary AirInjection) controller 1002, an FC (Fuel Cut) controller 1004, a deceleration leaningcontroller 1005, and an air-fuel ratio controller 1006 that functions as the air-fuelratio control system 10 in this embodiment as shown in Fig. 3.

The Al controller 1002 drives the secondary air pulse induction system 1000when a predetermined condition for taking in secondary air comes into effect, andtakes air from the air cleaner 34 and injects it as secondary air upstream of thecatalytic device 50 in the exhaust pipe 32. The Al controller 1002 outputs an Alinitiation signal Sais prior to the initiation of the injection of secondary air or whenthe injection is initiated, and outputs an Al termination signal Saie when theinjection of secondary air is finished.

The FC controller 1004 executes fuel cut control that interrupts the injectionof fuel when a predetermined condition for such FC control, that a throttle openingis turned to zero (the throttle valve is fully closed), comes into effect. The FCcontroller 1004 outputs an FC control initiation signal Sfcs prior to the initiation ofthe execution of fuel cut control or when the execution is initiated, and outputs anFC control termination signal Sfce when the fuel cut control is finished.

The deceleration leaning controller 1005 executes deceleration leaning suchthat basic injection pulse width is decreased when a predetermined condition fordeceleration leaning based upon the decrement of a throttle opening and thevariation of intake pressure comes into effect. The deceleration leaning controller1005 outputs a deceleration leaning initiation signal Srss prior to the initiation of the execution of deceleration leaning or when the execution is initiated, and outputs adeceleration leaning termination signal Srse when the deceleration leaning isfinished.

The air-fuel ratio controller 1006 is provided with a main body 100 of the air-fuel ratio controller, a basic fuel injection quantity calculating unit 116, and a first airintake corresponding part 1008A. The main body 100 of the air-fuel ratio controlleris a main body of the controller using sliding mode control and will be describedlater.

The basic fuel injection quantity calculating unit 116 calculates basic fuelinjection quantity TIMB by calculating a reference fuel injection quantity specifiedbased upon engine speed NE, a throttle opening TH and intake air pressure PBusing a basic fuel injection map 118, and correcting the reference fuel injectionquantity according to the effective opening area of the throttle valve 38.

The basic fuel injection map 118 is provided with a first basic fuel injectionmap 118a based upon engine speed NE and a throttle opening TH, and a secondbasic fuel injection map 118b based upon the engine speed NE and intake airpressure PB. Accordingly, in the air-fuel ratio controller 1006, there is provided amap selector 142 that selects and indicates the basic fuel injection map to be usedout of the first basic fuel injection map 118a and the second basic fuel injection map118b, based upon engine speed NE and the throttle opening TH from a map forselection 140 in which indexes of the basic fuel injection map to be used arearrayed. In the map for selection 140, as shown in Fig. 7, an area in which the firstbasic fuel injection map 118a is to be used and an area in which the second basicfuel injection map 118b is to be used are arranged. The map selector 142 selectsthe basic fuel injection map to be used based upon input engine speed NE and aninput throttle opening TH from the map for selection 140, and outputs a result of theselection Sa. When the engine speed NE is low, the probability that the first basicfuel injection map 118a is selected is high, and when the engine speed NE is high,the probability that the second basic fuel injection map 118b is selected is high.

Accordingly, the basic fuel injection quantity calculating unit 116 calculatesthe reference fuel injection quantity TIMB by calculating a reference fuel injectionquantity specified based upon engine speed NE, a throttle opening TH and intakeair pressure PB using the basic fuel injection map selected by the map selector142, and correcting the reference fuel injection quantity according to the effectiveopening area of the throttle valve 38. The basic fuel injection quantity TIMB is % corrected with target air-fuel ratio KO2(k) from the main body 100 of the air-fuel j ratio controller and an environmental correction factor KECO based on coolant Γ itemperature, intake air temperature and atmospheric pressure and others, and isoutput as fuel injection time Tout.

The first air intake corresponding part 1008A includes an air-fuel ratiocontrol halt requesting unit 1010, a rich spike controller 1012 and a resumptiondetermining unit 1014.

The air-fuel ratio control halt requesting unit 1010 outputs a halt requestsignal Sg to the main body 100 of the air-fuel ratio controller, based on the input ofan Al initiation signal Sais from the Al controller 1002 or an FC control initiationsignal Sfcs from the FC controller 1004 or a deceleration leaning initiation signalSrss from the deceleration leaning controller 1005. A controller 126 in the mainbody 100 of the air-fuel ratio controller temporarily halts air-fuel ratio control basedupon the input of a halt request signal Sg from the air-fuel ratio control haltrequesting unit 1010.

The rich spike controller 1012 controls the fuel injection valve 40 basedupon the input of an Al termination signal Saie from the Al controller 1002 or an FCcontrol termination signal Sfce from the FC controller 1004 or a deceleration leaningtermination signal Srse from the deceleration leaning controller 1005, andgenerates a rich spike (that is, temporarily performs rich injection with an air-fuelratio at which fuel injection quantity is richer than normal, so as to reduce NOXabsorbed in the catalytic device) in a combustion chamber of the engine 28. In therich spike control, fuel supply quantity according to a current output value (SVO2) ofthe oxygen sensor 52 is supplied to the combustion chamber of the engine 28 inunit time, referring to a rich spike map 1016 in which fuel supply quantity per unittime according to the output value (SVO2) of the oxygen sensor 52 is registered.The rich spike controller 1012 terminates the rich spike control when the outputvalue (SVO2) of the oxygen sensor 52 transfers to the rich side.

The resumption determining unit 1014 outputs a control resumption signal Sito the main body 100 of the air-fuel ratio controller when the output value (SVO2) ofthe oxygen sensor 52 transfers to the rich side. The main body 100 of the air-fuelratio controller resumes air-fuel ratio control based upon the input of the controlresumption signal Si. The resumption of air-fuel ratio control in the main body 100of the air-fuel ratio controller will be described later.

Next, the main body 100 of the air-fuel ratio controller will be describedreferring to Figs. 4 to 18.

The main body 100 of the air-fuel ratio controller is provided with a predictor j 102 (air-fuel ratio predicting means) that predicts air-fuel ratio downstream of thecatalytic device 50, a first sliding mode controller 104 (correction factor calculatingmeans) that determines a first correction factor DKO2OP(k) for fuel injectionquantity based on the predicted air-fuel ratio DVPRE from the predictor 102, anidentifier 106 that identifies parameters of the first sliding mode controller 104 andthe predictor 102, and an air-fuel ratio reference value calculating unit 108 thatcalculates an air-fuel ratio reference value as shown in Fig. 4.

The operation of the predictor 102, the first sliding mode controller 104, theidentifier 106 and the air-fuel ratio reference value calculating unit 108 will bedescribed in comparison with a comparative prior art example shown in Fig. 5 (amain body 300 of an air-fuel ratio controller similar to the air-fuel ratio controlsystem disclosed in JP 3373724).

First, the main body 300 of the air-fuel ratio controller in the comparativeexample shown in Fig. 5 assumes that an LAF sensor 110 (see a block shown by abroken line in Fig. 2) is installed upstream of the catalytic device 50, and the air-fuelratio A/F(k) before catalysis is input from the LAF sensor 110.

The predictor 102 predicts dead time dt (air-fuel ratio (VO2) after dead timeelapses corresponding to distance from the fuel injection valve 40 to the oxygensensor 52) from current time (k), so as to determine fuel injection quantity (targetair-fuel ratio) downstream of the catalytic device 50.

In a predictive model by the predictor 102, output V0Ut(k+dt)=Vpre(k) at thetime of k+dt can be predicted from the following relational expression (1) if only air-fuel ratio Φ in before catalysis between time ta and time tb and the output Vout of theoxygen sensor 52 are acquired as shown in Fig. 6 when current time is k.

[Mathematical expression 1]

However, as " Φ in" of "j = 1 ~(dt-d-1)" cannot be observed at the time of k, adesired value (Φορ) is used in place of it. In this case, V0U1'(K) shows deviation

between the output of the oxygen sensor 52 at the time of k and a desired value,and Vout'(K-1) shows deviation between the output of the oxygen sensor 52 beforethe time of k by one unit time (a fixed time cycle) and a desired value. "a1", "a2"and "Pj" are parameters determined by the identifier 106.

The first sliding mode controller 104 calculates injection quantity accordingto an error (predicted air-fuel ratio - desired value) of the model. Normally, slidingmode control, the concept of which is shown in Fig. 7, is a variable structure typefeedback control method in which a switching line represented by a linear functionhaving plural properties which are objects of control as variables is constructedbeforehand, the properties are converged on the switching line at high speed byhigh gain control (an arrival mode), and further, the properties are converged atrequired equilibrium points (converged points) on the switching line (a slidingmode), converging the properties on the switching line by so-called equivalent L control input.

Such sliding mode control has an excellent characteristic that the propertiescan be stably converged at the equilibrium points on the switching line withoutbeing substantially influenced by disturbance and others once the plural propertieswhich are the objects of control are converged on the switching line.

When a correction factor of air-fuel ratio of the engine 28 is calculated so as to stabilize the content of a specific component such as the oxygen content ofexhaust gas downstream of the catalytic device 50 at a predetermined optimumvalue, the correction factor of air-fuel ratio is calculated so as to converge a value ofthe content of the specific component of exhaust gas downstream of the catalyticdevice 50 and its varying velocity respectively as properties of an exhaust system tobe controlled for example at equilibrium points on the switching line (the points atwhich the value of the content and its varying velocity become predeterminedoptimum values and zero) using sliding mode control. When the correction factor ofair-fuel ratio is acquired using the sliding mode control, the content of the specificcomponent of exhaust gas downstream of the catalyst can be more preciselystabilized at the predetermined optimum value, compared with conventional typePID control and others. A switching function and a control input arithmetic expression in the slidingmode control are as follows.[Mathematical expression 2][Switching function]

[Control input arithmetic expression]

Equivalent rule input

: acquired from a conditional expression of

Arrival rule input

Adaptation rule input

In this case, Ueq(k) denotes equivalent rule input, Urch(k) denotes arrival ruleinput, Uadp(k) denotes adaptation rule input, and they are calculated in the followingexpression. Besides, in this case, Vout'(k) and Vout'(k-1) denote errors of the model,VOut'(k) denotes deviation between predicted air-fuel ratio at the time of k and adesired value, and Vout'(k-1) denotes deviation between predicted air-fuel ratiobefore the time of k by one unit time (a fixed time cycle) and a desired value.

Krch and Kadp denote feedback gain and S denotes a switching functionsetting parameter.

The identifier 106 compensates predictive precision in the predictor 102 bycorrecting a model parameter of the predictor 102. Besides, the identifier adjustsparameters a1(k), a2(k) and b1(k) in the first sliding mode controller 104 so as tominimize the deviation of Vout'(k+1) calculated in the following model expression byadjusting the convergence velocity (feedback gain) to the switching line of o(k)according to the error of the model.

[Mathematical expression 3]

As a result, correlation between air-fuel ratio before catalysis Φ in and Vout fortarget air-fuel ratio Φορ is corrected by correcting the model parameters in thepredictive expression.

The air-fuel ratio reference value calculating unit 108 calculates the air-fuelratio reference value of the engine 28 specified based upon adaptation rule inputUadp(k) from the first sliding mode controller 104 using a preset map.

Output from the first sliding mode controller 104, that is, control input Uop(=DKO2OP(k)) to the exhaust system, is added to the air-fuel ratio reference valuefrom the air-fuel ratio reference value calculating unit 108 in an adder 112, and thetarget air-fuel ratio KO2(k) is acquired. The target air-fuel ratio KO2(k) is input to anadaptive controller 114 at the next stage. The adaptive controller 114 is arecurrence formula type controller that adaptively calculates a feedback correctionfactor KAF based upon air-fuel ratio Φ,η (=A/F(k)) detected by the LAF sensor 110and target air-fuel ratio Φορ (KO2(k)) in consideration of a change of operational { status of the engine 28 and dynamic variation such as the variation of a property.

The basic fuel injection quantity calculating unit 116 acquires a referencefuel injection quantity specified based upon engine speed NE, a throttle opening THand intake air pressure PB using the preset basic fuel injection map 118, andcalculates basic fuel injection quantity TIMB by correcting the reference fuelinjection quantity according to the effective opening area of the throttle valve. Thebasic fuel injection quantity TIMB is supplied to a multiplier 120, is corrected by afeedback correction factor KAF from the adaptive controller 114 and an Ϊ

environmental correction factor KECO based upon coolant temperature, intake airtemperature, atmospheric pressure and others, and is output as fuel injection timeTout.

The main body 300 of the air-fuel ratio controller in the comparative examplehas a problem in that the expensive LAF sensor 110 is used, and so it cannot beapplied to a motorcycle or other vehicle that has a limit in terms of the costreduction of the system and space for arrangement. Then, in the main body 300 ofthe air-fuel ratio controller in the comparative example, as air-fuel ratio beforecatalysis Φίη cannot be measured when no LAF sensor 110 is installed upstream ofthe catalytic device 50, the predictive precision of air-fuel ratio after catalysis maybe deteriorated, the correction factor cannot be properly calculated when a greatlag with theoretical air-fuel ratio is caused because of the dispersion ofcharacteristics, the age deterioration and others of the engine 28 and the fuelinjection valve 40, and it is estimated that it is difficult to optimize air-fuel ratio.

Thus, the main body 100 of the air-fuel ratio controller in this embodiment isprovided with an adaptive model corrector 122 (adaptive model correcting means)that superimposes a second correction factor KTIMB on the first correction factorDKO2OP(k) so as to make a predictive error ERPRE(k) which is deviation betweenreal air-fuel ratio SVO2(k) and predicted air-fuel ratio DVPRE(k-dt) zero, a secondsliding mode/PID controller 124 that feedbacks so that an error between the realair-fuel ratio SVO2(k) and a preset desired value is made zero when predictiveprecision in the predictor 102 is deteriorated, a controller 126 that controls at leastthe first sliding mode controller 104 and the adaptive model corrector 122, and aswitching arrangement 128 that switches between output on the side of the firstsliding mode controller 104 and output on the side of the second sliding mode/PIDcontroller 124 based upon an instruction from the controller 126 as shown in Fig. 4.The switching arrangement 128 normally selects output on the side of the firstsliding mode controller 104, and switches to output on the side of the second slidingmode/PID controller 124 based upon a switching instruction signal Sd from thecontroller 126.

The second sliding mode/PID controller 124 is provided with a secondsliding mode control unit and a PID control unit, and one control unit is selectedaccording to an instruction from the controller 126. In normal operation (except fora special situation such as the injection of secondary air, fuel cut control anddeceleration leaning), one of the second sliding mode control unit or the PID control unit is selected according to the difference between an average of the transfer ofthe predictive error ERPRE(k) after filtering and a preset threshold. When thedifference is greater than a preset selection reference value, the PID control unit isselected, and when the difference is equal to or smaller than the selection referencevalue, the second sliding mode control unit is selected. The selection referencevalue may also be determined as follows. That is, a degree of convergence(convergence time and others) on the difference by the second sliding mode controlunit is grasped in an experiment and simulation beforehand; further, a range of thedifference in which the degree of convergence has no effect on another controlsystem is grasped, and the selection reference value is determined in the graspedrange of the difference.

As the second sliding mode control unit is selected when the difference isequal to or smaller than the selection reference value and feedback control is madeso as to make an error between real air-fuel ratio and a preset desired value zero,predictive precision can be secured at an early stage. Especially, as the PIDcontrol unit is selected when the difference exceeds the selection reference value,time until predictive precision is secured can be reduced more even if the differenceis great.

Besides, the controller 126 selects the PID control unit based upon thetermination of secondary air injection or the termination of fuel cut control or theinput of a control resumption signal Si output from the resumption determining unit1014 based upon the termination of deceleration leaning. Thus, even if thedifference is great, time until predictive precision is secured can be more reduced.That is, interrupted sliding mode control in the main body 100 of the air-fuel ratiocontroller can be restored at an early stage and emission performance can be | maintained at a high level.

Further, the main body 100 of the air-fuel ratio controller is provided with atiming device 130 that delays predicted air-fuel ratio DVPRE(k) from the predictor102 by dead time dt, and a subtractor 132 that differentiates output DVPRE(k-dt)from the timing device 130 and real air-fuel ratio SVO2(k) from the oxygen sensor52 to be predictive error ERPRE(k), and the predictive error ERPRE(k) from thesubtractor 132 is supplied to the adaptive model corrector 122. '1' is incremented tothe second correction factor KTIMB output from the adaptive model corrector 122by an adder 134. The output of the adder 134 and the target air-fuel ratio KO2(k)are multiplied by a multiplier 136, and the multiplied value is output as correction air-fuel ratio acquired by superimposing the second correction factor KTIMB on the j target air-fuel ratio KO2(k). The air-fuel ratio reference value is subtracted from the [ correction air-fuel ratio by a subtractor 138, and the subtracted value is input to thepredictor 102 and the identifier 106.

The adaptive model corrector 122 is provided with a filtering unit 144 thatfirst applies various filtering to a predictive error ERPRE(k), a predictive precisiondetermining unit 146 (predictive precision determining means) that determinespredictive precision based upon the filtered predictive error ERPRE(k), a firstcorrected quantity arithmetic unit 148a and a first correction factor arithmetic unit150a respectively corresponding to the first basic fuel injection map 118a, and asecond corrected quantity arithmetic unit 148b and a second correction factorarithmetic unit 150b respectively corresponding to the second basic fuel injectionmap 118b as shown in Fig. 8.

The first corrected quantity arithmetic unit 148a feedbacks predictive errorcorrected quantity 0th(i, j) so that a predictive error ERPRE(k) in which a weightcomponent based upon engine speed NE and a throttle opening TH is reflected ismade zero at a fixed time cycle when the first basic fuel injection map 118a isselected by the map selector 142. For example, operation is started before time kby dead time dt, that is, at time (k-dt), the operation is performed at the fixed timecycle, and at the time k, predictive error corrected quantity ©thlJ(k) is output.

Specifically, as shown in Fig. 9, the first corrected quantity arithmetic unit isprovided with a weighting unit 152 that superimposes a first weighting componentWSO2S(k) in which sensitivity for air-fuel ratio detected by the oxygen sensor 52 isreflected, a second weighting component Wtha(k-dt) in which the variation of valuesof the first basic fuel injection map 118a for the variation of engine speed NE and athrottle opening TH is reflected, and a third weighting component WthlJ(k-dt) inwhich the first basic fuel injection map 118a is made to correspond to plural areassorted based upon the engine speed NE and the throttle opening TH, on thepredictive error ERPRE(k) at the fixed time cycle and acquires a correction modelerror EwlJ(k) corresponding to the plural areas, and a sliding mode control unit 154that feedbacks respective predictive error corrected quantity 0thl J(k) correspondingto the plural areas so that correction model errors EwlJ(k) corresponding to theplural areas are made zero at the fixed time cycle.

To explain the first weighting component WSO2S(k), the output Vout of theoxygen sensor 52 has a nonlinear property for air-fuel ratio A/F as shown in Fig. IOA. In areas Za and Zc, even if air-fuel ratio varies, the output Vout of the oxygensensor 52 hardly varies. In the meantime, in an area Zb, the output Vout of theoxygen sensor 52 varies greatly according to a slight variation of air-fuel ratio A/F.In Fig. 10A, a full line La shows a property of a new catalyst and a broken line Lbshows a property of the catalyst after age deterioration. When such properties arereflected in the correction model error EwlJ(k) as they are, there is a problem that arapid change in the area Zb is input to the sliding mode control unit 154 and it takesmuch time to make the correction model error EwlJ(k) zero. Then, as shown in Fig. IOB, a value of weighting is reduced so that the rapid change is softened in thearea Zb.

To explain the second weighting component Wtha, as to an output valueSV02 of the oxygen sensor 52, the greater the inclination of basic fuel injectionquantity Tibs for the variation of a throttle opening TH is, the higher the probabilitythat a predictive error ERPRE is caused by an error in the detection of the throttleopening TH is, as shown in Fig. 11 A. When the error is caused in the detection anda reference point of a value of basic fuel injection quantity in the basic fuel injectionmap shifts, the variation of air-fuel ratio increases as “a value acquired by dividingvariation accompanied with the shift by the reference point” increases. Then, “avalue acquired by dividing the inclination of the basic fuel injection quantity Tibs forthe variation of a throttle opening TH by a value of the basic fuel injection quantityTibs” is set as each engine speed NE. As a result, as shown in Fig. 11B, whenengine speed NE is high, the second weighting component Wtha is substantially thesame even if a throttle opening TH is fully closed or is fully open, however, asengine speed NE gets low, the second weighting component Wtha increases as thethrottle opening TH decreases.

The third weighting component WthlJ is a function such that a weightingvalue linearly decreases toward an adjacent vertex from each vertex of enginespeed NE in the cases of 1000, 2000, 3000, 4500 (rpm) as to weighting functionsfor their engine speed as shown in Fig. 12A, for example. However, in Fig. 12A,when engine speed is 1000 rpm or less and is 4500 rpm or more, the weightingvalue is fixed. Similarly, as shown in Fig. 12B, weighting functions when a throttleopening TH is 1 °, 3°, 5°, 8° are such that a weighting value linearly decreasestoward an adjacent vertex from each vertex of these throttle openings TH.However, in Fig. 12B, when the throttle opening is 1° or less and is 8° or more, theweighting value is fixed.

The third weighting component WthlJ is acquired by multiplying weightingWthn(i) in engine speed NE and weighting Wtht(j) in a throttle opening TH.

The sliding mode control unit 154 feedbacks predictive error correctedquantity OthlJ for an area in which the third weighting component WthlJ is largerthan zero so that a correction model error EwIJ is zero and keeps the predictiveerror corrected quantity 0th IJ unupdated by making control input zero for an area inwhich the third weighting component WthlJ is zero.

The first correction factor arithmetic unit 150a acquires correction factorsKTITHIJ corresponding to plural areas by superimposing the third weightingcomponents WthlJ corresponding to the plural areas on predictive error correctedquantity OthlJ(k) corresponding to the plural areas at predetermined timing, adds allthe correction factors, and acquires a second correction factor KTIMB. In this case,as all the correction factors are added, the third weighting component WthlJ showsweighting according to a position of the following point in an area including the pointdetermined based upon engine speed NE and a throttle opening TH in the firstbasic fuel injection map 118a. Accordingly, as shown in Fig. 13, plural areas aremade by lattice points formed by 1000, 2000, 3000, 4500 (rpm) as engine speedand Γ, 3°, 5°, 8° as the throttle opening and when a point determined by inputengine speed NE and an input throttle opening TH is a point A, a correction factorcorresponding to the point A is interpolated by correction factors at four pointsaround the point A.

In the meantime, the second corrected quantity arithmetic unit 148bfeedbacks predictive error corrected quantity at the fixed time cycle so that apredictive error in which a weight component based upon engine speed NE andintake air pressure PB is reflected is zero when the second basic fuel injection map118b is selected by the map selector 142. For example, before time k by dead timedt, that is, at time (k-dt), operation is started, the operation is performed at the fixedtime cycle, and at the time k, predictive error corrected quantity ©pblJ(k) is output.As the specific configuration of the second corrected quantity arithmetic unit 148b issubstantially the same as that of the first corrected quantity arithmetic unit 148ashown in Fig. 9, the description is omitted.

The second correction factor arithmetic unit 150b acquires correction factorscorresponding to plural areas by superimposing the third weighting componentscorresponding to the plural areas on predictive error corrected quantity OpblJ(k)corresponding to the plural areas at predetermined timing, adds all the correction / factors, and acquires a second correction factor KTIMB. As the specificconfiguration of the second correction factor arithmetic unit 150b is alsosubstantially the same as that of the first correction factor arithmetic unit 150ashown in Fig. 9, the description is omitted.

The predictive precision determining unit 146 judges that predictiveprecision is deteriorated when a state in which the average of the transfer of thepredictive error ERPRE(k) after filtering is larger than a preset threshold continuesmore than a set frequency, and outputs a predictive precision deterioration signalSb. The predictive precision determining unit judges that predictive precision issecured when a state in which the average of the transfer of the predictive errorafter filtering is equal to or smaller than the preset threshold continues more thanthe set frequency, and outputs a predictive precision securement signal Sc. Thepredictive precision deterioration signal Sb and the predictive precision securementsignal Sc are supplied to the controller 126.

The controller 126 temporarily halts processing by the first sliding modecontroller 104 based on input of the predictive precision deterioration signal Sb (seeFig. 8) as shown in Fig. 4, temporarily halts the identifier 106, and reduces a cyclefor activating the adaptive model corrector 122 the while. That is, a fixed time cyclefor activating the first corrected quantity arithmetic unit 148a and the secondcorrected quantity arithmetic unit 148b is reduced. | i.

Besides, the controller 126 outputs a switching instruction signal Sd to theswitching arrangement 128 based on input of the predictive precision deteriorationsignal Sb. The switching arrangement 128 switches to output on the side of thesecond sliding mode/PID controller 124 based upon the input of the switchinginstruction signal Sd. The controller 126 also instructs the second sliding mode/PIDcontroller 124 to start processing based upon the input of the predictive precisiondeterioration signal Sb. In this case, predicted air-fuel ratio from the predictor 102is not used. In addition, the controller 126 selects either of the second sliding modecontrol unit or the PID control unit according to difference between the average ofthe transfer of the predictive error ERPRE(k) after filtering and the preset thresholdas described above. When the difference is greater than the preset selectionreference value, the PID control unit is selected, and when the difference is equal toor smaller than the selection reference value, the second sliding mode control unitis selected. Especially, the controller 126 forcibly selects the PID control unit basedupon the input of a control resumption signal Si output from the resumption determining unit 1014 according to the termination of secondary air injection or thetermination of fuel cut control or the termination of deceleration leaning. Thesecond sliding mode/PID controller 124 feedbacks so that an error between real air-fuel ratio (SVO2) and a preset desired value (for example, a fixed value showing anarea of a stoichiometric amount of air) is zero. Output from the second slidingmode/PID controller 124 is supplied to the multiplier 120 via the switchingarrangement 128. The basic fuel injection quantity calculating unit 116 calculatesreference injection quantity specified based upon engine speed NE, a throttleopening TH and intake air pressure PB using the preset basic fuel injection map orthe basic fuel injection map selected by the map selector 142, and calculates basicfuel injection quantity TIMB by correcting the reference fuel injection quantityaccording to the effective opening area of the throttle valve 38. The basic fuelinjection quantity TIMB is corrected based upon output (target air-fuel ratio KO2(k))from the switching arrangement 128 and the environmental correction factor KECOdetermined based upon coolant temperature, intake air temperature, atmosphericpressure and others, and is output as fuel injection time Tout.

The temporary halt of the first sliding mode controller 104 and the identifier106 may be also released by the output of a predictive precision securement signalSc by the predictive precision determining unit 146, and may be also released aftera predetermined time (time in which the securement of predictive precision isexpected) elapses. In this case, as the supply of a switching instruction signal Sdfrom the controller 126 to the switching arrangement 128 is stopped, the switchingarrangement 128 switches to output on the side of the first sliding mode controller104. Besides, the controller 126 restores the fixed time cycle for activating the firstcorrected quantity arithmetic unit 148a and the second corrected quantity arithmeticunit 148b in the adaptive model corrector 122. Moreover, the controller 126releases the temporary halt of the first sliding mode controller 104 and initializes aparameter of the identifier 106.

As described above, in the air-fuel ratio control system 10 (the air-fuel ratiocontroller 1006) in this embodiment, a value acquired by subtracting the air-fuelratio reference value from a value acquired by superimposing the second correctionfactor KTIMB on the target air-fuel ratio KO2(k) is input to the predictor 102 and theidentifier 106. That is, as predicted air-fuel ratio DVPRE(k) after dead time dt isoutput based upon the real air-fuel ratio SVO2(k) from the predictor 102, differencebetween the real air-fuel ratio SVO2(k) and the predicted air-fuel ratio DVPRE(k-dt) respectively temporally matched is input to the adaptive model corrector 122 as apredictive error ERPRE(k) by delaying the predicted air-fuel ratio DVPRE(k) by thedead time dt. The adaptive model corrector 122 superimposes the secondcorrection factor KTIMB on the first correction factor DKO2OP(k) so that thepredictive error ERPRE(k) is zero, the value is input to the predictor 102 and theidentifier 106, and the value is reflected in processing in the predictor 102.

That is, the first correction factor DKO2OP(k) acquired by feedbacking sothat deviation between predicted air-fuel ratio DVPRE(k) from the predictor 102 andtarget air-fuel ratio KO2(k) is zero and the second correction factor KTIMB acquiredby feedbacking so that the predictive error ERPRE(k) is zero are superimposed andits value is input to the predictor 102. Therefore, as the predictive precision of air-fuel ratio downstream of the catalytic device 50 can be secured even if the LAFsensor 110 heretofore installed upstream of the catalytic device 50 is removed, theair-fuel ratio of exhaust gas downstream of the catalytic device 50 can beconverged on an optimum value, and as a result, the decontamination performanceof the catalytic device 50 can be secured. Besides, even if the dispersion of theproperties of the engine 28, the fuel injection valve 40 and others and an error ofair-fuel ratio due to age deterioration and others are caused, the deterioration ofpredictive precision can be avoided. As described above, as the LAF sensor 110can be omitted, a harness and an interface circuit of the ECU 62 respectivelyrelated to the LAF sensor 110 can be omitted, the cost of the system and the spacefor arrangement can be reduced, and the air-fuel ratio control system according tothe present invention can be also readily applied to a vehicle having only little spacefor arrangement such as a motorcycle 12.

Normally, the LAF sensor 110 has to be maintained at a fixed temperatureby a heater so as to secure satisfactory operating characteristics; however, in thisembodiment, as the heater for the LAF sensor 110 can be also omitted, powerconsumption can be reduced and fuel economy can be enhanced.

Further, in this embodiment, as processing by the first sliding modecontroller 104 is temporarily halted based upon the input of a predictive precisiondeterioration signal Sb, the constraint of a cycle in the adaptive model corrector 122can be removed and the fixed time cycle for activating the first corrected quantityarithmetic unit 148a and the second corrected quantity arithmetic unit 148b can bereduced. Therefore, time until the predictive error ERPRE(k) is settled to zero canbe reduced. i

Furthermore, as processing in the second sliding mode/PID controller 124 isstarted based on input of a predictive precision deterioration signal Sb without usingpredicted air-fuel ratio DVPRE(k) from the predictor 102, fuel injection quantity iscontrolled so that real air-fuel ratio SVO2(k) approaches a predetermined desiredvalue and predictive precision can be secured in short time.

In the following cases (a) to (c), air-fuel ratio downstream of the catalyticdevice 50 can be also converged on an optimum value at an early stage by suchprocessing operation, and the deterioration of emission caused by the continuationof a state in which the air-fuel ratio of exhaust gas downstream of the catalyticdevice 50 cannot be converged on the optimum value can be avoided. i (a) The case that as an air-fuel ratio error is caused due to the dispersion ofthe properties of the engine 28, the fuel injection valve 40 and others and agedeterioration, a great predictive error that exceeds a range which can be adjustedby the predictor 102 is identified by the identifier 106 (b) The case that dynamic characteristics to be controlled rapidly vary (thevariation in volume of exhaust gas by a change of an operating condition, the use ofethanol mixed fuel and others) (c) The case that the oxygen sensor 52 is located in a dead zone (an area inwhich the output of the oxygen sensor 52 hardly varies even if air-fuel ratio varies)

Furthermore, in this embodiment, as at a stage at which it is determined thatpredictive precision is secured, the cycle for activating the adaptive model corrector122 is restored and the temporary halt of the first sliding mode controller 104 isreleased, the generation of the first correction factor DKO2OP(k) by the first slidingmode controller 104 is resumed at a stage at which predictive precision is secured,the predictive precision is further enhanced, and the optimization of air-fuel ratiodownstream of the catalytic device 50 can be accelerated.

In this case, as parameters of the identifier 106 are initialized, thesecurement of predictive precision can be maintained by using initial values asidentification parameters when predictive precision is secured or at a stage at whichthe securement of predictive precision is expected without using the identificationparameters when predictive precision is deteriorated, and the optimization of air-fuel ratio downstream of the catalytic device 50 can be accelerated.

Furthermore, as the first corrected quantity arithmetic unit 148a of theadaptive model corrector 122 feedbacks predictive error corrected quantity OthlJ atthe fixed time cycle so that a predictive error in which a weight component based upon engine speed NE and a throttle opening TH for the first basic fuel injectionmap 118a is reflected is zero and the second correction factor KTIMB is calculatedbased upon the predictive error corrected quantity ©th IJ at predetermined timing inthe first correction factor arithmetic unit 150a, air-fuel ratio downstream of thecatalytic device 50 can be optimized even if the LAF sensor 110 upstream of thecatalytic device 50 is removed.

Especially, as predictive error corrected quantity OthlJ corresponding toplural area is feedbacked at the fixed time cycle so that the respective predictiveerror corrected quantity OthlJ corresponding to the plural areas sorted in the firstbasic fuel injection map 118a based upon engine speed NE and a throttle openingTH is zero, correction factors KTITHIJ corresponding to the plural areas arecalculated based upon the predictive error corrected quantity OthlJ corresponding tothe plural areas at predetermined timing, all the correction factors are added andthe second correction factor KTIMB is acquired, the second correction factor KTIMBis a value acquired by correcting used map values using the correction factorsKTITHIJ of the plural areas so that a predictive error ERPRE(k) is zero.Accordingly, when the second correction factor KTIMB having such a characteristicis superimposed on the first correction factor DKO2OP, air-fuel ratio downstream ofthe catalytic device 50 can be optimized.

This is also similar as to the second corrected quantity arithmetic unit 148aand the second correction factor arithmetic unit 150b respectively corresponding tothe second basic fuel injection map 118b.

In the above-mentioned example, the processing of the first sliding modecontroller 104 and the identifier 106 is temporarily halted at the stage at which thedeterioration of predictive precision is determined and the switching arrangement128 switches to output from the second sliding mode/PID controller 124; however,for example, the processing of the first sliding mode controller 104 and the identifier106 is temporarily halted based upon the input of a signal Se telling that an air-fuelratio feedback condition from the ECU 62 comes into effect and the switchingarrangement 128 may also switch to output from the second sliding mode/PIDcontroller 124. In this case, when a predictive error is caused due to an operatingcondition or similar before the air-fuel ratio feedback condition comes into effect, thepredictive error can be eliminated at an initial stage after the air-fuel ratio feedbackcondition comes into effect. After a predetermined time (a time in which thesecurement of predictive precision is expected) elapses after the input of the signal

Se telling that the air-fuel ratio feedback condition comes into effect, the temporaryhalt may be also released.

Furthermore, as the cycle for activating the adaptive model corrector 122 isrestored at a stage at which the preset time (the predetermined time) elapses afterthe deterioration of predictive precision is determined and the generation of the firstcorrection factor DKO2OP(k) by the first sliding mode controller 104 is resumed at astage at which the predictive precision is secured after the predetermined timeelapses once or more when the temporary halt of the first sliding mode controller104 is released, the predictive precision is further enhanced and the optimization ofair-fuel ratio downstream of the catalytic device 50 can be accelerated. Predictiveprecision is secured by setting predetermined time for one time as a time in whichthe securement of the predictive precision is expected when predetermined time forat least two times elapses.

Furthermore, when operation gain according to the correction factor of theadaptive model corrector 122 is set to be larger than normal gain, the similar effectcan be also acquired in place of temporarily halting the processing of the first slidingmode controller 104 and the identifier 106 and reducing the cycle for activating theadaptive model corrector 122.

Next, the alternative embodiments of the above-mentioned main body 100of the air-fuel ratio controller will be described referring to Figs. 14 to 18. A main body 100a of an air-fuel ratio controller in a first alternativeembodiment as shown in Fig. 14 has a substantially similar configuration to that ofthe main body 100 of the air-fuel ratio controller; however, the main body 100a isdifferent in that target air-fuel ratio KO2(k) from an adder 112 and a secondcorrection factor KTIMB from an adaptive model corrector 122 are added in anadder 160. In this case, a value acquired by adding a first correction factorDKO2OP(k) and the second correction factor KTIMB is also input to a predictor 102and an identifier 106. Accordingly, a similar effect to the main body 100 of the air-fuel ratio control system can be acquired. A main body 100b of an air-fuel ratio controller in a second alternativeembodiment as shown in Fig. 15 has a substantially similar configuration to that ofthe main body 100 of the air-fuel ratio controller; however, the main body 100b inthe second variation is different in that target air-fuel ratio KO2(k) is acquired bymultiplying a value (KO2OP(k)) acquired by adding output from an adder 112 (avalue (KO2OP(k) acquired by adding a first correction factor DKO2OP(k) from a first sliding mode controller 104 and an air-fuel ratio reference value from an air-fuelratio reference value calculating unit 108) and output (a value acquired byincrementing a second correction factor KTIMB by 1) from an adder 134 in an adder162, without reflecting the second correction factor KTIMB in a predictor 102 and anidentifier 106. In this case, as the second correction factor KTIMB is reflected in theoutput of a basic fuel injection quantity calculating unit 116, a similar effect to themain body 100 of the air-fuel ratio controller in this embodiment can be acquired. A main body 100c of an air-fuel ratio controller in a third alternativeembodiment as shown in Fig. 16 has a substantially similar configuration to that ofthe main body 100b of the air-fuel ratio controller in the second alternativeembodiment; however, the main body 100c is different in that target air-fuel ratioKO2(k) is acquired by adding output KO2SL(k) from an adder 112 and a secondcorrection factor KTIMB from an adaptive model corrector 122 in an adder 164. Inthis case, as the second correction factor KTIMB is reflected in the output of a basicinjection quantity calculating unit 116, a similar effect to the main body 100 of theair-fuel ratio controller in this embodiment can be also acquired. A main body 100d of an air-fuel ratio controller in a fourth alternativeembodiment as shown in Fig. 17 has a substantially similar configuration to that ofthe main body 100 of the air-fuel ratio controller in this embodiment; however, a firstswitching arrangement 128a is installed between a predictor 102 and a first slidingmode controller 104, and a second switching arrangement 128b is installed on theoutput side of the first sliding mode controller 104. Normally, the predictor 102 isselected by the first switching arrangement 128a and output to the adder 112 isselected by the second switching arrangement 128b. Hereby, as predicted air-fuelratio DVPRE(k) from the predictor 102 is input to the first sliding mode controller104, a first correction factor DKO2OP(k) from the first sliding mode controller 104 isadded to an air-fuel ratio reference value in the adder 112 and is output as targetair-fuel ratio KO2(k). In the meantime, when a switching instruction signal Sd isoutput from a controller 126, the first switching arrangement 128a selects the inputof real air-fuel ratio SVO2(k) and the second switching arrangement 128b selectsoutput to a multiplier 120. Hereby, the first sliding mode controller 104 feedbacksso that an error between the real air-fuel ratio (SVO2) and a preset desired value(for example, a fixed value showing an area of a stoichiometric amount of air) iszero. This output from the first sliding mode controller 104 is supplied to themultiplier 120 via the second switching arrangement 128b. Accordingly, in the fourth alternative embodiment, a similar effect to the main body 100 of the air-fuelratio controller in this embodiment can be also acquired. Especially, according tothe fourth alternative embodiment, a second sliding mode/PID controller 124 can beomitted and the configuration can be simplified. A main body 100e of an air-fuel ratio controller in a fifth alternativeembodiment as shown in Fig. 18 has substantially the same configuration as that ofthe main body 100 of the air-fuel ratio controller in this embodiment; however, themain body 100e is different in that an LAF sensor 110 is installed upstream of acatalytic device 50, and the detected air-fuel ratio A/F(k) from the LAF sensor 110 isutilized. In this case, an adaptive controller 114 is installed between a switchingarrangement 128 and a multiplier 120.

The deterioration of predictive precision caused by the shortage of theprecision of a basic fuel injection map can be settled at an early stage by utilizingthe LAF sensor 110. Naturally, as the main body 100 of the air-fuel ratio controllerin this embodiment, the main body 100a of the air-fuel ratio controller in the firstalternative embodiment to the main body 100d of the air-fuel ratio controller in thefourth alternative embodiment, as the first correction factor DKO2OP(k) from thefirst sliding mode controller 104 and the second correction factor KTIMB from theadaptive model corrector 122 are superimposed and are input to the predictor 102and the identifier 106, the deterioration of predictive precision can be settled at anearly stage, however, the deterioration of predictive precision caused by theshortage of the precision of the basic fuel injection map 118 can be settled at anearly stage by utilizing the LAF sensor 110.

The main body 100 of the air-fuel ratio controller in this embodiment and thevarious alternatives can be applied not only to the air-fuel ratio control of an engine,but also to a control system in which delay in transportation since control input tilloutput is long and a predictor 102 is required to be configured.

Next, the processing operation of the first air intake corresponding part1008A will be described with reference to a flowchart shown in Fig. 19.

First, in a step S1 shown in Fig. 19, the air-fuel ratio control halt requestingunit 1010 determines whether the injection of secondary air is started or not. Thisdetermination is made depending upon whether an Al initiation signal Sais is inputfrom the Al controller 1002 or not.

When the Al initiation signal Sais is input, the process proceeds to the nextstep S2, and the air-fuel ratio control halt requesting unit 1010 outputs a halt requesting signal Sg to the main body 100 of the air-fuel ratio controller. The mainbody 100 of the air-fuel ratio controller temporarily halts air-fuel ratio control basedupon the input of the halt requesting signal Sg.

Afterward, in a step S3, the rich spike controller 1012 waits for thetermination of the injection of secondary air. That is, the rich spike controller waitsuntil an Al termination signal Saie is input from the Ai controller 1002.

In the meantime, if it is determined in the step S1 that no Al initiation signalSais is input, the air-fuel ratio control halt requesting unit 1010 determines whetherFC control is started or not in a step S4. This determination is made dependingupon whether an FC control initiation signal Sfcs is input from the FC controller1004 or not.

When the FC control initiation signal Sfcs is input, the process proceeds tothe next step S5 and the air-fuel ratio control halt requesting unit 1010 outputs ahalt requesting signal Sg to the main body 100 of the air-fuel ratio controller. Themain body 100 of the air-fuel ratio controller temporarily halts air-fuel ratio controlbased upon the input of the halt requesting signal Sg.

Afterward, in a step S6, the rich spike controller 1012 waits for thetermination of the FC control. That is, the rich spike controller waits for the input ofan FC control termination signal Sfce from the FC controller 1004.

If it is determined in the step S4, that no FC control initiation signal Sfcs isinput, the air-fuel ratio control halt requesting unit 1010 determines whetherdeceleration leaning is started or not in a step S7. This determination is madedepending upon whether a deceleration leaning initiation signal Srss is input fromthe deceleration leaning controller 1005 or not.

When the deceleration leaning initiation signal Srss is input, the processproceeds to the next step S8 and the air-fuel ratio control halt requesting unit 1010outputs a halt request signal Sg to the main body 100 of the air-fuel ratio controller.The main body 100 of the air-fuel ratio controller temporarily halts air-fuel ratiocontrol based upon the input of the halt request signal Sg.

Afterward, in a step S9, the rich spike controller 1012 waits for thetermination of deceleration leaning. That is, the rich spike controller waits for theinput of a deceleration leaning termination signal Srse from the deceleration leaningcontroller 1005.

When the Al termination signal Saie is input in the step S3 or when the FCcontrol termination signal Sfce is input in the step S6 or when the deceleration leaning termination signal Srse is input in the step S9, the process proceeds to thenext step S10, the rich spike controller 1012 controls the fuel injection valve 40, andthe rich spike controller applies rich spike control to the combustion chamber of theengine 28. In the rich spike control, the supplied quantity of fuel according to anoutput value (SVO2) of the oxygen sensor 52 at the current time is supplied to thecombustion chamber of the engine 28 in unit time, referring to the rich spike map1016. The rich spike control is made until the output value (SVO2) of the oxygensensor 52 transfers to the rich side.

In a step S11, the resumption determining unit 1014 determines whether theoutput value (SVO2) of the oxygen sensor 52 has transferred to the rich side or not,and outputs a control resumption signal Si to the main body 100 of the air-fuel ratiocontroller at a stage at which the output value (SVO2) of the oxygen sensor 52transfers on the rich side.

In a step S12, the main body 100 of the air-fuel ratio controller resumes air-fuel ratio control based upon the input of the control resumption signal Si.

The processing in steps S1 to S12 will be described based upon a timingchart shown in Fig. 20. Fig. 20 shows an execution period Ta of the injection ofsecondary air or FC control or deceleration leaning, a rich spike control period Tb (aresidual oxygen processing period of the catalytic device 50), the variation of fuelinjection quantity per unit time in rich spike control, the variation (as a full line) of thequantity of oxygen stored in the catalytic device 50 and the variation (as a brokenline) of the output value (SVO2) of the oxygen sensor.

First, at a time point t1, when the injection of secondary air (Al) or FC controlor deceleration leaning is started, the real air-fuel ratio SVO2 shown by an outputvalue of the oxygen sensor 52 transfers to the lean side and as a result, thequantity of oxygen stored in the catalytic device 50 gradually increases.

At a stage at which the quantity of oxygen stored in the catalyst reaches amaximum, the injection of secondary air (Al) or FC control or deceleration leaning isfinished and at a termination time point t2, rich spike control, that is, the residualoxygen processing of the catalytic device 50, is performed. In the rich spike control,between the time point t2 and a time point t3, normally, fuel at a level to provide avery rich air-fuel ratio for a stoichiometric amount of air in a combustible area isinjected (a high-concentration injection period Tb1). Accordingly, in the high-concentration injection period Tb1, the quantity of oxygen stored in the catalyticdevice 50 rapidly decreases. At a stage at which an output value SVO2 of the oxygen sensor 52 starts to gradually increase toward a stoichiometric level,injection quantity per unit time is gradually reduced. Between the time point t3 anda time point t4, normally, fuel at a level to provide a richer air-fuel ratio than thestoichiometric level in a range that has no effect upon output is injected (a low-concentration injection period Tb2). Accordingly, in the low-concentration injectionperiod Tb2, the quantity of oxygen stored in the catalytic device 50 decreases moreslowly than that in the high-concentration injection period Tb1. At a stage at whichthe output value SVO2 of the oxygen sensor 52 gradually approaches thestoichiometric level, fuel injection quantity per the unit time is gradually reduced.

At a stage at which an output value SVO2 of the oxygen sensor 52 transfersto the rich side, rich spike control is finished and feedback control in the main body100 of the air-fuel ratio controller is resumed.

The predictive precision determining unit 146 (see Fig. 8) in the main body100 of the air-fuel ratio controller that resumes control determines whetherpredictive precision by the predictor 102 is deteriorated or not in a step S13 shownin Fig. 19. That is, the predictive precision determining unit determines whether astate in which an average of the transfer of the predictive error ERPRE(k) afterfiltering is larger than the preset threshold continues more than the set frequency ornot. When predictive precision by the predictor 102 is deteriorated, the processproceeds to a step S14, the controller 126 temporarily halts control in the firstsliding mode controller 104, and the controller selects the PID control unit in thesecond sliding mode/PID controller 124. At this time, the switching arrangement128 switches to output on the side of the second sliding mode/PID controller 124according to the supply of a switching instruction signal Sd from the controller 126.The PID control unit feedbacks so that an error between real air-fuel ratio (SVO2)and the preset desired value is zero. Hereby, output from the second slidingmode/PID controller 124 is supplied to the multiplier 120 via the switchingarrangement 128. Hereby, the basic fuel injection quantity TIMB is corrected withoutput from the second sliding mode/PID controller 124 (in this case, the PIDcontrol unit) and the environmental correction factor KECO based upon coolanttemperature, intake air temperature, atmospheric pressure and others, is output asfuel injection time Tout, and real air-fuel ratio SVO2 varies toward the desired value.

In the step S13, at a stage at which it is determined that predictive precisionis secured, the process proceeds to a step S15 and the controller 126 resumescontrol in the first sliding mode controller 104. That is, the controller releases the temporary halt of the first sliding mode controller 104 and initializes the parameterof the identifier 106. Besides, the controller halts the supply of a switchinginstruction signal Sd to the switching arrangement 128. The switching arrangement128 switches to output on the side of the first sliding mode controller 104. Hereby,the basic fuel injection quantity TIMB is corrected with output KO2(k) from the firstsliding mode controller 104 and the environmental correction factor KECO basedupon coolant temperature, intake air temperature, atmospheric pressure and othersand is output as fuel injection time Tout.

Afterward, at a stage at which feedback control in the first sliding modecontroller 104 is resumed in the step S15 or when it is determined that decelerationleaning is not started in the step S7 (that is, when the injection of secondary air orfuel cut control or deceleration leaning is not made), the process proceeds to a stepS16 and it is determined whether a request for the termination (the disconnection ofpower supply, a maintenance request and others) of the air-fuel ratio control system10 is made or not. When no request for termination is made, processing after thestep S1 is repeated and at a stage at which a request for termination is made,processing operation in the air-fuel ratio control system 10 is finished.

As described above, in the air-fuel ratio control system 10 in thisembodiment, even if it is a system in which the secondary air pulse inductionsystem 1000 is installed, the quantity of oxygen stored in the catalyst by theinjection of secondary air can be reduced at an early stage by interrupting slidingmode control in the main body 100 of the air-fuel ratio controller after the injection ofsecondary air and executing rich injection control at a stage at which the injection ofsecondary air is finished. As a result, the interrupted sliding mode control isrestored at an early stage and emission performance can be kept high.

When the halt of fuel injection is controlled by the FC controller 1004 byclosing the throttle opening, the quantity of oxygen stored in the catalytic device 50increases because air containing oxygen itself flows into the catalyst. However, inthe air-fuel ratio control system 10 in this embodiment, as sliding mode control inthe main body 100 of the air-fuel ratio controller is interrupted after fuel injection haltcontrol by the FC controller 1004 and rich injection control is executed at a stage atwhich the fuel injection halt control is finished, the quantity of oxygen stored in thecatalytic device 50 by the intake of air containing oxygen according to the fuelinjection halt control can be reduced at an early stage. As a result, the interrupted sliding mode control can be restored at an early stage and emission performance i can be kept high.

Besides, as air containing oxygen itself flows into the catalyst as in the caseof the fuel cut control when deceleration leaning is started, the quantity of oxygenstored in the catalyst increases. However, according to the air-fuel ratio controlsystem 10 in this embodiment, as feedback control is interrupted in decelerationleaning by the deceleration leaning controller 1005 and rich injection control isperformed at a stage at which the deceleration leaning is finished, the quantity ofoxygen stored in the catalyst by the intake of air containing oxygen according to thedeceleration leaning can be reduced at an early stage. As a result, the interruptedfeedback control is restored at an early stage and emission performance can bekept high.

Moreover, when it is predicted that time is required for convergence onsliding mode control by the injection of secondary air or fuel cut control or the intakeof air containing oxygen according to the deceleration leaning, convergence onfeedback control can be accelerated by first performing PID control at a stage atwhich the injection of secondary air is finished or at a stage at which fuel cut controlor deceleration leaning is finished.

As the convergence on sliding mode control is secured when a predictiveerror is equal to or smaller than the threshold, emission performance can be kepthigh by resuming feedback control.

Especially, as PID control is made so that an error between real air-fuel ratioand the preset desired value is zero without using the predictor 102 when thedeterioration of predictive precision is determined in the resumption of control in themain body 100 of the air-fuel ratio controller, time until predictive precision issecured can be more reduced than time when the predictor 102 is used andemission performance can be kept high.

Next, a second air intake corresponding part 1008B will be describedreferring to Figs. 21 and 22.

The second air intake corresponding part 1008B as shown in Fig. 21 has asubstantially similar configuration to that of the first air intake corresponding part1008A; however, the second air intake corresponding part 1008B is different fromthe first air intake corresponding part 1008A in that an error arithmetic unit 1018, aPID controller 1020 and a switching arrangement 1022 are further provided.Besides, processing in the following units included in the second air intake corresponding part 1008B is partially different from the processing in the air-fuelratio control halt requesting unit 1010 and the resumption determining unit 1014.

That is, the air-fuel ratio control halt requesting unit 1010 outputs a haltrequest signal Sg to the main body 100 of the air-fuel ratio controller based uponthe input of an Al initiation signal Sais from the Al controller 1002 or an FC controlinitiation signal Sfcs from the FC controller 1004 or a deceleration leaning initiationsignal Srss from the deceleration leaning controller 1005, and outputs a firstswitching signal Sh1 to the switching arrangement 1022. The main body 100 of theair-fuel ratio controller temporarily halts air-fuel ratio control based upon the input ofthe halt request signal Sg from the air-fuel ratio control halt requesting unit 1010.Besides, the switching arrangement 1022 switches to output from the PID controller1020 based upon the input of the first switching signal Sh1.

The error arithmetic unit 1018 operates an error ERR between an outputvalue (SVO2) of the oxygen sensor 52 at the current time and the preset desiredvalue.

The PID controller 1020 executes PID control (feedback control) so that theerror ERR acquired in the error arithmetic unit 1018 is zero. The output of the PIDcontroller 1020 is input to the multiplier 120 via the switching arrangement 1022 astarget air-fuel ratio KO2(k). Hereby, the basic fuel injection quantity TIMB iscorrected with the output from the PID controller 1020 and the environmentalcorrection factor KECO based upon coolant temperature, intake air temperature,atmospheric pressure and others, is output as fuel injection time Tout, and anoutput value SVO2 of the oxygen sensor 52 varies toward the desired value.

The resumption determining unit 1014 compares the error ERR from theerror arithmetic unit 1018 and the preset threshold, outputs a control resumptionsignal Si to the main body 100 of the air-fuel ratio controller at a stage at which theerror ERR is equal to or smaller than the threshold, and outputs a second switchingsignal Sh2 to the switching arrangement 1022. The threshold may be the same asthe threshold used in the predictive precision determining unit 146 in the main body100 of the air-fuel ratio controller. The main body 100 of the air-fuel ratio controllerresumes air-fuel ratio control based upon the input of the control resumption signalSi from the resumption determining unit 1014 and the switching arrangement 1022switches to output from the main body 100 of the air-fuel ratio controller based uponthe input of the second switching signal Sh2. Hereby, output from the main body100 of the air-fuel ratio controller is input to the multiplier 120 via the switching arrangement 1022 as target air-fuel ratio KO2(k). That is, the basic fuel injectionquantity TIMB is corrected with the output from the main body 100 of the air-fuelratio controller and the environmental correction factor KECO based upon coolanttemperature, intake air temperature, atmospheric pressure and others, and isoutput as fuel injection time Tout.

Next, processing operation of the second air intake corresponding part 1008B will be described with reference to a flowchart shown in Fig. 22.

First, in a step S101 shown in Fig. 22, the air-fuel ratio control haltrequesting unit 1010 determines whether the injection of secondary air is started ornot. This determination is made depending upon whether an Al initiation signalSais from the Al controller 1002 is input or not.

When the Al initiation signal Sais is input, the process proceeds to the nextstep S102, the air-fuel ratio control halt requesting unit 1010 outputs a halt requestsignal Sg to the main body 100 of the air-fuel ratio controller, and the air-fuel ratiocontrol halt requesting unit outputs a first switching signal Sh1 to the switchingarrangement 1022. The main body 100 of the air-fuel ratio controller temporarilyhalts air-fuel ratio control based upon the input of the halt request signal Sg. Theswitching arrangement 1022 switches to output from the PID controller 1020 basedupon the input of the first switching signal Sh1.

Afterward, in a step S103, the rich spike controller 1012 waits for the\ termination of the injection of secondary air. That is, the rich spike controller waits\ for the input of an Al termination signal Saie from the Al controller 1002.

In the meantime, when it is determined that no Al initiation signal Sais is

input in the step S101, the air-fuel ratio control halt requesting unit 1010 determineswhether FC control is started or not in a step S104. This determination is madeI depending upon whether an FC control initiation signal Sfcs is input from the FCcontroller 1004 or not.

When the FC control initiation signal Sfcs is input, the process proceeds tothe next step S105, the air-fuel ratio control halt requesting unit 1010 outputs a haltrequest signal Sg to the main body 100 of the air-fuel ratio controller, and the air-fuel ratio control halt requesting unit outputs a first switching signal Sh1 to theswitching arrangement 1022.

Afterward, in a step S106, the rich spike controller 1012 waits for thetermination of FC control. That is, the rich spike controller waits for the input of anFC control termination signal Sfce from the FC controller 1004.

When it is determined that no FC control initiation signal Sfcs is input in thestep S104, the air-fuel ratio control halt requesting unit 1010 determines whetherdeceleration leaning is started or not in a step S107. This determination is madedepending upon whether a deceleration leaning initiation signal Srss from thedeceleration leaning controller 1005 is input or not.

When the deceleration leaning initiation signal Srss is input, the processproceeds to the next step S108 and the air-fuel ratio control halt requesting unit1010 outputs a halt request signal Sg to the main body 100 of the air-fuel ratiocontroller. The main body 100 of the air-fuel ratio controller temporarily halts air-fuel ratio control based upon the input of the halt request signal Sg.

Afterward, in a step S109, the rich spike controller 1012 waits for thetermination of deceleration leaning. That is, the rich spike controller waits for theinput of a deceleration leaning termination signal Srse from the deceleration leaningcontroller 1005.

When the Al termination signal Saie is input in the step S103 or when theFC control termination signal Sfce is input in the step S106 or when thedeceleration leaning termination signal Srse is input in the step S109, the processproceeds to the next step S110 and the rich spike controller 1012 controls the fuelinjection valve 40 to generate a rich spike in the combustion chamber of the engine28. The rich spike control supplies fuel of quantity according to an output value(SVO2) of the oxygen sensor 52 at the current time to the combustion chamber ofthe engine 28 in unit time, referring to the rich spike map 1016. The rich spikecontrol is continued until it is determined in a step S111 that an output value(SVO2) of the oxygen sensor 52 has transferred to the rich side.

At a stage at which the output value of the oxygen sensor 52 transfer to therich side, the process proceeds to a step S112 and the error arithmetic unit 1018calculates difference (an error) between an output value SVO2 of the oxygensensor 52 at the current time and the desired value.

Afterward, in a step S113, the resumption determining unit 1014 determineswhether control in the main body 100 of the air-fuel ratio controller is to be resumedor not. This determination is made depending upon whether the error is equal to orsmaller than the threshold or not. When the error exceeds the threshold, theprocess proceeds to a step S114 and the PID controller 1020 performs PID controlso that the error is zero. The output of the PID controller 1020 is input to themultiplier 120 via the switching arrangement 1022 as target air-fuel ratio KO2(k).

Hereby, the basic fuel injection quantity TIMB is corrected with the output from thePID controller 1020 and the environmental correction factor KECO based uponcoolant temperature, intake air temperature, atmospheric pressure and others, isoutput as fuel injection time Tout, and an output value SVO2 of the oxygen sensor52 varies toward the desired value.

In the step S113, at a stage at which it is determined that the error is equalto or smaller than the threshold, the process proceeds to the next step S115, theresumption determining unit 1014 outputs a control resumption signal Si to the mainbody 100 of the air-fuel ratio controller, and outputs a second switching signal Sh2to the switching arrangement 1022. The main body 100 of the air-fuel ratiocontroller resumes air-fuel ratio control based upon the input of the controlresumption signal Si from the resumption determining unit 1014 and the switchingarrangement 1022 switches to output from the main body 100 of the air-fuel ratiocontroller based upon the input of the second switching signal Sh2. Hereby, theoutput from the main body 100 of the air-fuel ratio controller is input to the multiplier120 via the switching arrangement 1022 as target air-fuel ratio KO2(k). That is, thebasic fuel injection quantity TIMB is corrected with the output from the main body100 of the air-fuel ratio controller and the environmental correction factor KECObased upon coolant temperature, intake air temperature, atmospheric pressure andothers, and is output as fuel injection time Tout.

Afterward, at a stage at which feedback control in the first sliding modecontroller 104 is resumed in the step S115 or when it is determined thatdeceleration leaning is not started in the step S107 (that is, when the injection ofsecondary air, fuel cut control or deceleration leaning is not started), the processproceeds to a step S116, and it is determined whether a request for the termination(the disconnection of power supply, a maintenance request and others) of the air-fuel ratio control system 10 is made or not. When no request for the termination ismade, the processing after the step S101 is repeated and at a stage at which therequest for termination is made, processing operation in the air-fuel ratio controlsystem 10 is finished.

As described above, in the air-fuel ratio control system 10 provided with thesecond air intake corresponding part 1008B, PID control is performed so thatdifference (an error) between an output value SVO2 of the oxygen sensor 52 andthe desired value is zero using the dedicated PID controller 1020 without using thesecond sliding mode/PID controller 124 in the main body 100 of the air-fuel ratio controller, and so the error can be converged on the threshold or less at an earlystage without using a predictive precision determination routine (from the predictor102 to the timing device 130, the filtering unit 144 in the adaptive model corrector122 and the predictive precision determining unit 146) using a predictive error in themain body 100 of the air-fuel ratio controller, normal sliding mode control (control inthe first sliding mode controller 104) can be resumed more early than in the caseusing the first air intake corresponding part 1008A, and emission performance canbe kept high.

Besides, the second air intake corresponding part 1008B can be alsoapplied to a configuration in which no second sliding mode/PID controller 124 isused as in the main body 100d of the air-fuel ratio controller in the fourth alternativeembodiment shown in Fig. 17, has high flexibility, and can be applied to variousspecifications.

Claims

1. An air-fuel ratio control system comprising: a basic fuel injection map (118) that specifies fuel injection quantity to anengine (28) based upon at least parameters of engine speed and throttle opening; air-fuel ratio sensing means (52) which is provided downstream of a catalyst (50) installed in an exhaust pipe (32) of the engine (28) and which senses air-fuelratio; air-fuel ratio predicting means (102) that predicts air-fuel ratio downstream of the catalyst (50); and correction factor calculating means (104) that determines a correction factor (DKO2OP) for the fuel injection quantity based upon predicted air-fuel ratio (DVPRE) from the air-fuel ratio predicting means (102) by feedback control, wherein: ! the air-fuel ratio predicting means (102) calculates the predicted air-fuel ratio (DVPRE) based upon at least real air-fuel ratio (SVO2) from the air-fuel ratio sensing means (52) and a history of the correction factor (DKO2OP); the air-fuel ratio control system has adaptive model correcting means (122) that superimposes a second correction factor (KTIMB) on the correction factor(DKO2OP) so as to make deviation between the real air-fuel ratio (SVO2) and thepredicted air-fuel ratio (DVPRE) which corresponds to the real air-fuel ratio (SVO2)and which is predicted in the past as a predictive error (ERPRE) zero; a secondary air pulse induction system (1000) that injects secondary air into an exhaust path upstream of the catalyst (50) is provided; the feedback control is interrupted at a stage at which the injection ofsecondary air is started; immediately after the injection of secondary air is finished, air-fuel ratio control is made to shift to rich control in an open loop; and after the real air-fuel ratio (SVO2) transfers to a rich side, the feedbackcontrol is resumed.

2. An air-fuel ratio control system comprising: a basic fuel injection map (118) that specifies fuel injection quantity to anengine (28) based upon at least parameters of engine speed and throttle opening; air-fuel ratio sensing means (52) which is provided downstream of a catalyst(50) installed in an exhaust pipe (32) of the engine (28) and which senses air-fuelratio; air-fuel ratio predicting means (102) that predicts air-fuel ratio downstreamof the catalyst (50); and correction factor calculating means (104) that determines a correction factor(DKO2OP) for the fuel injection quantity based upon predicted air-fuel ratio(DVPRE) from the air-fuel ratio predicting means (102) by feedback control, wherein: the air-fuel ratio predicting means (102) calculates the predicted air-fuel ratio(DVPRE) based upon at least real air-fuel ratio (SVO2) from the air-fuel ratiosensing means (52) and a history of the correction factor (DKO2OP); the air-fuel ratio control system has adaptive model correcting means (122)that superimposes a second correction factor (KTIMB) on the correction factor(DKO2OP) so as to make deviation between the real air-fuel ratio (SVO2) and thepredicted air-fuel ratio (DVPRE) which corresponds to the real air-fuel ratio (SVO2)and which is predicted in the past as a predictive error (ERPRE) zero; the air-fuel ratio control system has fuel cut control means (1004) thatperforms fuel injection halt control while the throttle opening is closed; when the fuel injection halt control by the fuel cut control means (1004) isstarted, the feedback control is interrupted; immediately after the fuel injection halt control by the fuel cut control means(1004) is finished, air-fuel ratio control is made to shift to rich control in an openloop; and after the real air-fuel ratio (SVO2) transfers to a rich side, the feedbackcontrol is resumed.

3. An air-fuel ratio control system comprising: a basic fuel injection map (118) that specifies fuel injection quantity to anengine (28) based upon at least parameters of engine speed and throttle opening; air-fuel ratio sensing means (52) which is provided downstream of a catalyst(50) installed in an exhaust pipe (32) of the engine (28) and which senses air-fuelratio; air-fuel ratio predicting means (102) that predicts air-fuel ratio downstreamof the catalyst (50); and correction factor calculating means (104) that determines a correction factor (DKO2OP) for the fuel injection quantity based upon predicted air-fuel ratio (DVPRE) from the air-fuel ratio predicting means (102) by feedback control, wherein: the air-fuel ratio predicting means (102) calculates the predicted air-fuel ratio (DVPRE) based upon at least real air-fuel ratio (SVO2) from the air-fuel ratiosensing means (52) and a history of the correction factor (DKO2OP); the air-fuel ratio control system has adaptive model correcting means (122) that superimposes a second correction factor (KTIMB) on the correction factor(DKO2OP) so as to make deviation between the real air-fuel ratio (SVO2) and thepredicted air-fuel ratio (DVPRE) which corresponds to the real air-fuel ratio (SVO2)and which is predicted in the past as a predictive error (ERPRE) zero; the air-fuel ratio control system has deceleration leaning control means (1005) that performs deceleration leaning; when the deceleration leaning by the deceleration leaning control means (1005) is started, the feedback control is interrupted; immediately after the deceleration leaning by the deceleration leaningcontrol means (1005) is finished, air-fuel ratio control is made to shift to rich controlin an open loop; and after the real air-fuel ratio (SVO2) transfers to a rich side, the feedback i control is resumed. j

4. The air-fuel ratio control system according to any one of Claims 1 to 3, wherein: the feedback control is sliding mode control; and when the predictive error (ERPRE) exceeds a preset threshold prior toresumption of the feedback control, PID control is performed so as to make an errorbetween the real air-fuel ratio (SVO2) and a preset desired value zero.

5. The air-fuel ratio control system according to Claim 4, wherein: the feedback control is resumed at a stage at which the predictive error(ERPRE) is equal to or smaller than the preset threshold.

6. The air-fuel ratio control system according to any one of Claims 1 to 3,comprising: a controller (126) that controls at least the correction factor calculatingmeans (104) and the adaptive model correcting means (122), wherein: the adaptive model correcting means (122) is provided with predictiveprecision determining means (146) that determines predictive precision based uponthe predictive error (ERPRE); and when deterioration of the predictive precision is determined in the predictiveprecision determining means (146) in the resumption of the feedback control, thePID control is performed without using the air-fuel ratio predicting means (102) soas to make the error between the real air-fuel ratio (SVO2) and the preset desiredvalue zero.

7. The air-fuel ratio control system according to any one of Claims 1 to 3,further comprising: a dedicated PID controller (1020), wherein: the feedback control is the sliding mode control; and when the error between the real air-fuel ratio (SVO2) and the preset desiredvalue exceeds the preset threshold prior to the resumption of the feedback control,the PID control is performed in the PID controller (1020) so as to make the errorzero. I