JP4240325B2 - Control device - Google Patents

Description

  The present invention relates to a control apparatus that controls a control input to be converged to a target value by calculating a control input to the control target based on a ΔΣ modulation algorithm or the like.

  Conventionally, as this type of control device, for example, one described in Patent Document 1 is known. This control device detects the output of the controlled object and outputs the detection result as an analog amount detection signal, and the deviation for calculating the deviation between the analog amount target value input from the host device and the detection signal Calculation means, conversion means for converting the calculated deviation into a 1-bit digital signal, and compensation means for compensating the 1-bit digital signal from the conversion means and outputting it as an operation signal are provided (FIG. 6 of the same publication). reference).

In this control device, the deviation (analog amount) between the detection signal and the target value is calculated by the deviation calculating means, and this calculated deviation is converted into a 1-bit digital signal by ΔΣ modulation by the converting means, and further by the compensating means. After being compensated, it is input to the control object as an operation signal. With the above configuration, an operation amount having a phase opposite to the deviation is generated and input to the control target so as to cancel the deviation between the output of the control target and the target value. As a result, feedback control is performed so that the output of the controlled object converges to the target value.

JP 2001-154704 A

  According to the above conventional control device, when the dynamic characteristic of the control target has a relatively large phase delay or dead time, the input signal is reflected after the input signal is input to the control target. It takes time until the output signal to be output from the control target, and a control timing shift occurs between the input and output of the control target. As a result, the control system may become unstable. For example, when the air-fuel ratio of the exhaust gas of the internal combustion engine is controlled using the fuel injection amount of the internal combustion engine as an input, it takes time from the actual fuel injection to the actual change of the air-fuel ratio of the exhaust gas. Therefore, the stability and controllability of the air-fuel ratio control is lowered, and the characteristics of the exhaust gas purified by the catalyst may become unstable.

  The present invention has been made to solve the above problem, and even when the controlled object exhibits a relatively large dynamic characteristic such as a phase delay or a dead time, the control timing deviation between the input and output of the controlled object is eliminated. It is an object of the present invention to provide a control device that can improve control stability and controllability.

In order to achieve this object, the control device 1 according to claim 1 is based on the prediction algorithm (Equations (6) and (7)) and outputs the control target (the output Vout of the oxygen concentration sensor 15) and a predetermined target. Predicted value calculating means (ECU2, state predictor 22, step 33) for calculating a predicted value PREVO2 of the output deviation VO2 , which is a deviation from the value Vop, and a correction coefficient (gain KRDSM, G _d ) Multiplying means (ECU 2, step 195) and correction for setting the correction coefficient to a smaller value when the predicted value of output deviation is greater than or equal to a predetermined value (value 0) than when it is less than the predetermined value coefficient setting means (ECU 2, step 192-194) and, delta modulation algorithm, one of the modulation algorithm of the ΔΣ modulation algorithm, and ΣΔ modulation algorithm The basis, the correction coefficient (gain KRDSM, G _d) in accordance with the predicted value PREVO2 of the output deviation is multiplied, the output of the controlled object of a given order to control so as to converge to the target value Vop, to the controlled object Control input calculating means (ECU2, ADSM controller 20, step 38) for calculating a control input (target air-fuel ratio KCMD).

According to this control device, based on the prediction algorithm, the predicted value of the output deviation, which is the deviation between the output of the controlled object and the predetermined target value, is calculated, and any one of the Δ modulation algorithm, the ΔΣ modulation algorithm, and the ΣΔ modulation algorithm Since the control input is calculated according to the predicted value of the output deviation multiplied by the correction coefficient based on the one modulation algorithm, such a predicted value is used as a dynamic characteristic such as a phase delay or a dead time of the control target. By calculating as a value reflecting the above, it is possible to eliminate a shift in control timing between the input and output of the control target. Further, a control input for controlling the output to be controlled to converge to a predetermined target value is calculated according to the predicted value of the output deviation multiplied by the correction coefficient, and this correction coefficient is calculated based on the output deviation. When the predicted value is equal to or greater than the predetermined value, the value is set to be smaller than when the predicted value is less than the predetermined value. Therefore, the output of the control target is converged to the target value according to the magnitude relationship of the predicted value of the output deviation with respect to the predetermined value. The convergence speed can be changed. As a result, control stability can be ensured and controllability can be improved (in this specification, “calculation” such as “calculation of predicted value” and “calculation of control input” is calculated by a program. Including, but not limited to, generating electrical signals representing them by hardware).

According to a second aspect of the present invention, in the control device 1 according to the first aspect, the predicted value calculating means is configured to calculate a control input (target air-fuel ratio KCMD) and a control input input to a control target based on a prediction algorithm. The output deviation prediction value PREVO2 is calculated in accordance with at least one of the values reflecting the above (the output KACT of the LAF sensor 14) and the control target output (the output Vout of the oxygen concentration sensor 15).

According to this control device, the predicted value of the output deviation can be calculated while reflecting the state of the control input, and the calculation accuracy (prediction accuracy) of the predicted value of the output deviation can be increased accordingly. As a result, control stability can be ensured and controllability can be improved.

According to a third aspect of the present invention, in the control device 1 according to the first or second aspect, the prediction algorithm is a value representing one of a control input and a value reflecting the control input input to the control target (air-fuel ratio deviation). It is an algorithm based on a controlled object model (formula (1)) using DKCMD, LAF output deviation DKACT) and output deviation VO2 as variables.

According to this control apparatus, since the predicted value is calculated based on the control target model using the value representing one of the control input or the value reflecting the control input and the output deviation as a variable, the control target model Is defined as a value that reflects the dynamic characteristics such as the phase delay and dead time of the controlled object, so that the predicted output deviation is a value that reflects the dynamic characteristics such as the phase delay and dead time of the controlled object. Can be calculated. In general, in the control target model, the deviation between the input / output of the control target and the specified value is defined as a variable that represents the input / output rather than the absolute value of the input / output defined as a variable. It is known that the parameters can be more accurately identified or defined so that the dynamic characteristics of the controlled object model can be adapted to the actual dynamic characteristics of the controlled object. Therefore, according to this control apparatus, since the control target model uses the output deviation, which is the deviation between the output of the control target and the predetermined target value, as a variable, compared with the case where the absolute value of the output of the control target is a variable. Thus, it is possible to improve the adaptability of the dynamic characteristic of the controlled object model to the actual dynamic characteristic of the controlled object, thereby further increasing the calculation accuracy of the predicted value of the output deviation. As a result, control stability can be ensured and controllability can be improved.

According to a fourth aspect of the present invention, in the control device 1 according to the third aspect , the value representing one of the control input and the value reflecting the control input input to the control target is the control input (target air-fuel ratio KCMD). And a predetermined reference value FLAFBASE (air-fuel ratio deviation DKCMD), and a value (LAF sensor 14 output KACT) reflecting a control input input to the control target and a predetermined reference value FLAFBASE (LAF output deviation) DKACT).

  As described above, in the control target model, the deviation between the input / output of the control target and the predetermined value is defined as a variable representing the input / output, rather than the absolute value of the input / output defined as a variable. The dynamic characteristics of the controlled object model can be adapted to the actual dynamic characteristics of the controlled object. Therefore, according to this control apparatus, the controlled object model calculates the deviation between the calculated control input and the predetermined reference value, or the difference between the value reflecting the control input input to the controlled object and the predetermined reference value. Since it is a variable, compared with the case where the absolute value of the control input or the value that reflects the control input is a variable, the suitability of the controlled model's dynamic characteristics to the actual dynamic characteristics of the controlled object can be improved, Thereby, the calculation accuracy of the predicted value can be further increased.

According to a fifth aspect of the present invention, in the control device 1 according to any one of the first to fourth aspects, the control input calculation means is multiplied by a correction coefficient (gain KRDSM, G _d ) based on one modulation algorithm . An intermediate value (DSM signal value DSMSGNS) is calculated according to the output deviation predicted value PREVO2 , and a control input (target air-fuel ratio KCMD, adaptive) is calculated based on a value obtained by multiplying the calculated intermediate value by a predetermined gain (KDSM). The target air-fuel ratio (KCMDLSD) is calculated.

  In general, each of the ΔΣ modulation algorithm, the ΣΔ modulation algorithm, and the Δ modulation algorithm determines the control input on the assumption that the gain to be controlled is 1, so that the actual gain to be controlled is different from the value 1. If the control input is not calculated to an appropriate value, the controllability may be reduced. For example, when the actual gain of the control target is larger than the value 1, the control input is calculated as a value larger than necessary, which may result in an overgain state. On the other hand, according to this control device, the control input is calculated based on a value obtained by multiplying the intermediate value calculated based on one modulation algorithm by a predetermined gain. By appropriately setting, good controllability can be ensured.

According to a sixth aspect of the present invention, in the control device 1 according to the fifth aspect , the gain parameter detecting means (ECU 2, the intake pipe absolute pressure sensor 11) for detecting the gain parameter (exhaust gas volume AB_SV) representing the gain characteristic of the controlled object. , Crank angle sensor 13, step 80) and gain setting means (ECU2, step 200) for setting a predetermined gain value (KDSM) in accordance with the detected gain parameter. .

  According to this control device, since the predetermined gain used for calculating the control input is set according to the gain characteristic of the control target, the control input is calculated as a value having appropriate energy according to the gain characteristic of the control target. Thus, the occurrence of an overgain state or the like can be avoided, and good controllability can be ensured.

According to a seventh aspect of the present invention, in the control device 1 according to any one of the first to sixth aspects, the control input calculating means is multiplied by a correction coefficient (gain KRDSM, G _d ) based on one modulation algorithm . A second intermediate value (ΔΣ modulation control amount DKCMDDSM) is calculated according to the output deviation predicted value PREVO2 (step 201), and a predetermined value (reference value FLAFBASE) is added to the calculated second intermediate value. Thus, the control input (adaptive target air-fuel ratio KCMDLSLD) is calculated (step 211).

  In general, any of the Δ modulation algorithm, the ΔΣ modulation algorithm, and the ΣΔ modulation algorithm can only calculate a positive / negative inversion type control input centered on a value of 0. On the other hand, according to this control apparatus, the control input calculation unit calculates the control input by adding a predetermined value to the second intermediate value calculated based on one modulation algorithm. The control input can be calculated not only as a value that reverses positive and negative around the value 0, but also as a value that repeatedly increases and decreases the predetermined width around the predetermined value, so that the degree of freedom of control can be increased.

According to an eighth aspect of the present invention, in the control device 1 according to any one of the first to seventh aspects, the predicted value calculating means outputs the control target after the control input (target air-fuel ratio KCMD) is input to the control target. The predicted time dt until it is reflected in (the output Vout of the oxygen concentration sensor 15) is calculated according to the dynamic characteristic (exhaust gas volume AB_SV) to be controlled (step 81), and according to the calculated predicted time dt. Thus, the output deviation predicted value PREVO2 is calculated (step 33).

According to this control device, the predicted time from when the control input is input to the controlled object until it is reflected in the output of the controlled object is calculated according to the dynamic characteristics of the controlled object, and the calculated predicted time is Accordingly, the predicted value of the output deviation is calculated accordingly, and by calculating the control input using the predicted value of the output deviation calculated in this way, the control caused by the response delay or dead time of the controlled object A shift in control timing between the input and output of the target can be eliminated more reliably, and the controllability can be further improved.

According to a ninth aspect of the present invention, in the control device 1 according to the second aspect, the output of the controlled object is downstream of the catalyst (first catalyst device 8a) in the exhaust passage (exhaust pipe 7) of the internal combustion engine 3. is arranged, the output Vout of the downstream air-fuel ratio sensor for detecting an air-fuel ratio of the exhaust gas after passing through the catalyst (oxygen concentration sensor 15), the braking control input to control the subject is supplied to the internal combustion engine 3 A value that is the target air-fuel ratio KCMD of the air-fuel mixture and that reflects the control input input to the control target is arranged upstream of the catalyst (first catalyst device 8a) in the exhaust passage, and the exhaust gas before passing through the catalyst This is the output KACT of the upstream air-fuel ratio sensor (LAF sensor 14) that detects the air-fuel ratio of the gas, and the predicted value calculation means is the target of the air-fuel mixture supplied to the internal combustion engine based on the prediction algorithm (formula (7)) Air-fuel ratio KCMD and upstream air And at least one output KACT ratio sensor, in accordance with an output Vout of the downstream air-fuel ratio sensor, calculates a predicted value PREVO2 of the output deviation VO2, the control input-calculating means, based on the one modulation algorithm, the correction coefficient ( The target air of the mixture to be supplied to the internal combustion engine for converging the output Vout of the downstream air-fuel ratio sensor to a predetermined target value Vop according to the predicted value of the output deviation multiplied by the gain KRDSM, G _d ). It is characterized by comprising air-fuel ratio calculating means (ECU2, step 38) for calculating the fuel ratio KCMD.

According to this control device, the predicted value of the output deviation, which is the deviation between the output of the downstream air-fuel ratio sensor and the predetermined target value , is the target air-fuel ratio and / or upstream air-fuel ratio of the air-fuel mixture supplied to the internal combustion engine. the output of the sensor is calculated in accordance with the output of the downstream air-fuel ratio sensor, the target air-fuel ratio of the mixture for converging the output of the downstream air-fuel ratio sensor to a predetermined target value, the correction coefficient is multiplied It is calculated based on one modulation algorithm according to the predicted value of the output deviation. Since the control input, that is, the target air-fuel ratio of the air-fuel mixture is calculated as described above, by appropriately setting a predetermined target value, the exhaust gas air-fuel ratio can be changed to a state in which the exhaust gas purification state by the catalyst is good. As a result, the characteristics of the exhaust gas purified by the catalyst (hereinafter referred to as “post-catalyst exhaust gas characteristics”) can be improved. Further, since the predicted value is calculated according to the output of the upstream air-fuel ratio sensor provided above the catalyst, the state of the air-fuel ratio of the exhaust gas actually supplied to the catalyst is appropriately reflected by the predicted value. Therefore, the calculation accuracy of the predicted value can be improved accordingly. Further, since the correction coefficient to be multiplied by the predicted value of output deviation is set to a smaller value when the predicted value of output deviation is greater than or equal to a predetermined value, it is set to a predetermined value of the predicted value of output deviation. The convergence speed for converging the output of the downstream air-fuel ratio sensor to a predetermined target value can be changed according to the magnitude relationship with respect to the value. Therefore, for example, when this predetermined value is set to the value 0, when the predicted value of the output deviation is 0 or more, that is, the output of the downstream air-fuel ratio sensor is larger than the target value, the target air-fuel ratio is set to the lean side. When the change is to be made, the convergence speed is set to be lower than when the change is made to the rich side, so that the effect of suppressing the NOx emission amount due to the lean bias can be obtained. On the other hand, when the target air-fuel ratio should be changed to the rich side, the NOx purification rate of the catalyst can be sufficiently recovered by setting the convergence speed to a value larger than that when changing to the lean side.

According to a tenth aspect of the present invention, in the control device 1 according to the ninth aspect , the operating state detecting means (ECU2, intake pipe absolute pressure) for detecting the operating state of the internal combustion engine (engine speed NE, intake pipe absolute pressure PBA). The sensor 11 and the crank angle sensor 13) are further provided, and the predicted value calculation means is provided with the downstream air-fuel ratio sensor after the air-fuel mixture of the target air-fuel ratio is supplied to the internal combustion engine according to the detected operating state of the internal combustion engine. A predicted time dt until the output is reflected is calculated, and a predicted value PREVO2 of the output deviation VO2 is further calculated in accordance with the calculated predicted time dt.

  In this type of control device that controls the air-fuel ratio, the dynamic characteristics (for example, response delay and dead time) of the controlled objects including the internal combustion engine and the catalyst vary according to the operating state of the internal combustion engine, for example, the exhaust gas volume. On the other hand, according to this control device, the predicted time from when the air-fuel mixture having the target air-fuel ratio is supplied to the internal combustion engine until it is reflected in the output of the downstream air-fuel ratio sensor is calculated according to the operating state. In addition, since the predicted value of the output deviation is calculated further in accordance with the calculated predicted time, the control input is calculated using the predicted value calculated in this way, so that the dynamic characteristics of the controlled object can be obtained. The resulting shift in the control timing between the input and output of the control target can be eliminated more reliably, and the post-catalyst exhaust gas characteristics can be further improved.

The invention according to claim 11 is the control device 1 according to claim 9 , further comprising operating state detecting means for detecting the operating state of the internal combustion engine, wherein the air-fuel ratio calculating means is based on one modulation algorithm, and the correction coefficient Intermediate value calculation means for calculating an intermediate value (DSM signal value DSMSGNS) of the target air-fuel ratio of the air-fuel mixture to be supplied to the internal combustion engine according to the output deviation predicted value PREVO2 multiplied by (gain KRDSM, G _d ). ECU2, steps 195 to 199), gain setting means (ECU2, step 200) for setting the gain (KDSM) according to the detected operating state of the internal combustion engine, and the gain set to the calculated intermediate value A target air-fuel ratio calculating method for calculating a target air-fuel ratio KCMD (adaptive target air-fuel ratio KCMDLSD) of the air-fuel mixture to be supplied to the internal combustion engine based on the multiplied value. (ECU 2, step 211) and, characterized in that it comprises a.

  In this type of control device that controls the air-fuel ratio, the gain characteristic with respect to the air-fuel ratio of the control target including the internal combustion engine and the catalyst changes according to the operating state of the internal combustion engine, for example, the exhaust gas volume. In this case, as described above, since one modulation algorithm determines the control input on the assumption that the gain to be controlled is a value of 1, if the gain characteristic of the control target changes as described above, The target air-fuel ratio of the air-fuel mixture becomes oscillating greatly away from an appropriate value, and the output of the downstream air-fuel ratio sensor downstream of the catalyst is also oscillating. As a result, the post-catalyst exhaust gas characteristics are deteriorated. In contrast, according to this control device, the target air-fuel ratio of the air-fuel mixture is calculated based on a value obtained by multiplying the intermediate value calculated based on one modulation algorithm by the gain, and the gain is calculated. Since it is set according to the operating state, the target air-fuel ratio of the air-fuel mixture can be calculated as a value that appropriately reflects the change in the gain characteristic of the controlled object that accompanies the change in the operating state. Can be further improved.

According to a twelfth aspect of the present invention, in the control device 1 according to the second aspect, the output to be controlled is disposed downstream of the catalyst (second catalyst device 8b) in the exhaust passage of the internal combustion engine and passes through the catalyst. the output Vout of the air-fuel ratio sensor (oxygen concentration sensor 15) for detecting an air-fuel ratio of the exhaust gas after, braking control input to the control object is located at the target air-fuel ratio KCMD of the mixture supplied to the internal combustion engine The predicted value calculation means calculates a predicted value PREVO2 of the output deviation according to the target air-fuel ratio of the air-fuel mixture supplied to the internal combustion engine and the output of the air-fuel ratio sensor based on the prediction algorithm (equation (6)), control input-calculating means, based on the one modulation algorithm, the correction coefficient (gain KRDSM, G _d) in accordance with the predicted value PREVO2 of the output deviation is multiplied, the air-fuel ratio predetermined target value output Vout of the sensor For focusing the op, characterized in that it consists of an air-fuel ratio calculating means for calculating a target air-fuel ratio of the mixture to be supplied to an internal combustion engine (ECU 2).

According to this control device, the predicted value of the output deviation, which is the deviation between the output of the air-fuel ratio sensor and the predetermined target value , depends on the target air-fuel ratio of the air-fuel mixture supplied to the internal combustion engine and the output of the air-fuel ratio sensor. The target air-fuel ratio of the air-fuel mixture for converging the output of the air-fuel ratio sensor to a predetermined target value in accordance with the predicted value of the output deviation multiplied by the correction coefficient is calculated based on one modulation algorithm. Calculated. Since the control input, that is, the air-fuel ratio of the air-fuel mixture is calculated as described above, by appropriately setting a predetermined target value, the exhaust gas air-fuel ratio is brought into a good state of purification of the exhaust gas by the catalyst. As a result, the post-catalyst exhaust gas characteristics can be improved. Further, since only a single air-fuel ratio sensor needs to be used, such a control device can be realized relatively inexpensively.

According to a thirteenth aspect of the present invention, in the control device 1 according to the twelfth aspect of the present invention, operating state detecting means (ECU2, intake pipe absolute pressure) for detecting the operating state of the internal combustion engine (engine speed NE, intake pipe absolute pressure PBA). The sensor 11 and the crank angle sensor 13) are further provided, and the predicted value calculation means outputs the air-fuel ratio sensor output after the air-fuel mixture of the target air-fuel ratio is supplied to the internal combustion engine according to the detected operating state of the internal combustion engine. A predicted time dt until reflection is calculated, and a predicted value PREVO2 of the output deviation is further calculated according to the calculated predicted time dt.

According to this control device, the same effect as that attained by the 10th aspect can be attained.

The invention according to claim 14, in the control apparatus 1 of claim 12, further comprising a driving state detection means for detecting an operating condition of the internal combustion engine, the air-fuel ratio calculating means, based on the one modulation algorithm, the correction factor Intermediate value calculation means (ECU2) for calculating the intermediate value (DSM signal value DSMSGNS) of the target air-fuel ratio of the air-fuel mixture to be supplied to the internal combustion engine according to the predicted value of the output deviation multiplied by (gain KRDSM, G _d ) ) And gain setting means (ECU2) for setting the gain (KDSM) according to the detected operating state of the internal combustion engine, and a value obtained by multiplying the calculated intermediate value by the gain set, And target air-fuel ratio calculating means (ECU2) for calculating a target air-fuel ratio KCMD (adaptive target air-fuel ratio KCMDLSD) of the air-fuel mixture to be supplied to the engine.

According to this control apparatus, the same effect as that attained by the 11th aspect can be attained.

  According to this control device, the predicted value of the value representing the output of the control target is calculated based on the prediction algorithm to which the control target model is applied, and the control input is calculated based on one modulation algorithm. Calculated according to the value. In this case, as described above, by using the model parameter identified by the identification unit, the dynamic characteristic of the controlled object model can be adapted to the actual dynamic characteristic of the controlled object. By calculating based on the applied prediction algorithm, the predicted value can be calculated as a value reflecting the actual dynamic characteristic of the controlled object. As a result, the control timing shift between the control input and the output of the control target can be corrected more appropriately, and the control stability and controllability can be further improved.

  According to a twenty-ninth aspect of the present invention, in the control device 1 according to the twenty-eighth aspect, the control input calculating means calculates the predicted time dt from when the control input is input to the control target until it is reflected in the output of the control target. It is calculated according to the dynamic characteristic (exhaust gas volume AB_SV) to be controlled (step 81), and based on the prediction algorithm, the predicted value PREVO2 is calculated according to the calculated predicted time dt (step 33). To do.

  According to this control device, the same effect as that attained by the 9th aspect can be attained.

  The invention according to claim 30 is the control device 1 according to any one of claims 18 to 29, wherein the control input calculation means calculates an intermediate value (DSM signal value DSMSGNS) based on the controlled object model and one modulation algorithm. The control input (target air-fuel ratio KCMD) is calculated based on a value obtained by multiplying the calculated intermediate value by a predetermined gain (KDSM).

  According to this control apparatus, it is possible to obtain the same effect as that of the sixth aspect of the invention.

  According to a thirty-first aspect of the present invention, in the control device 1 according to the thirty-third aspect, gain parameter detecting means (ECU2, intake pipe absolute pressure sensor 11) for detecting a gain parameter (exhaust gas volume AB_SV) representing a gain characteristic to be controlled. , Crank angle sensor 13, step 80) and gain setting means (ECU2, step 200) for setting a predetermined gain value (KDSM) in accordance with the detected gain parameter. .

  According to this control device, it is possible to obtain the same effect as that of the seventh aspect of the invention.

  According to a thirty-second aspect of the present invention, in the control device 1 according to any one of the eighteenth to thirty-first aspects, the control input calculating means is based on one modulation algorithm, and the second intermediate value (ΔΣ modulation) according to the predicted value. (Control amount DKCMDDSM) is calculated (step 201), and a predetermined value (reference value FLAFBASE) is added to the calculated second intermediate value to calculate a control input (adaptive target air-fuel ratio KCMDLSLD) (step 211). ).

  As described above, according to the control device of the present invention, even when the controlled object exhibits a relatively large dynamic characteristic such as a phase delay or a dead time, the control timing deviation between the input and output of the controlled object is eliminated. Thereby, the stability and controllability of the control can be improved.

  Hereinafter, a control device according to a first embodiment of the present invention will be described with reference to the drawings. The first embodiment is an example in which the control device is configured to control the air-fuel ratio of the internal combustion engine, and FIG. 1 shows a schematic configuration of the control device 1 and the internal combustion engine 3 to which the control device 1 is applied. As shown in the figure, the control device 1 includes an ECU 2. As will be described later, the ECU 2 mixes to be supplied to the internal combustion engine (hereinafter referred to as "engine") 3 according to the operating state. Control the air / fuel ratio of the air.

  The engine 3 is an in-line four-cylinder gasoline engine mounted on a vehicle (not shown) and includes first to fourth cylinders # 1 to # 4. In the vicinity of the throttle valve 5 of the intake pipe 4 of the engine 3, a throttle valve opening sensor 10 constituted by, for example, a potentiometer is provided. The throttle valve opening sensor 10 detects the opening θTH of the throttle valve 5 (hereinafter referred to as “throttle valve opening”) θTH and sends the detection signal to the ECU 2.

  Further, an intake pipe absolute pressure sensor 11 is provided downstream of the throttle valve 5 of the intake pipe 4. The intake pipe absolute pressure sensor 11 (gain parameter detection means, operating state detection means, dynamic characteristic parameter detection means) is constituted by, for example, a semiconductor pressure sensor, and detects the intake pipe absolute pressure PBA in the intake pipe 4. A detection signal is output to the ECU 2.

  The intake pipe 4 is connected to four cylinders # 1 to # 4 via four branch portions 4b of the intake manifold 4a. An injector 6 is attached to each branch portion 4b upstream of an intake port (not shown) of each cylinder. When each engine 6 is operated, the final fuel injection amount TOUT that is the valve opening time and the injection timing are controlled by the drive signal from the ECU 2.

  On the other hand, a water temperature sensor 12 composed of, for example, a thermistor is attached to the main body of the engine 3. The water temperature sensor 12 detects the engine water temperature TW, which is the temperature of the cooling water circulating in the cylinder block of the engine 3, and outputs a detection signal to the ECU 2.

  A crank angle sensor 13 is provided on the crankshaft (not shown) of the engine 3. The crank angle sensor 13 (gain parameter detecting means, operating state detecting means, dynamic characteristic parameter detecting means) outputs a CRK signal and a TDC signal, which are pulse signals, to the ECU 2 as the crankshaft rotates.

  One pulse of the CRK signal is output every predetermined crank angle (for example, 30 °). The ECU 2 calculates the engine speed (hereinafter referred to as “engine speed”) NE of the engine 3 in accordance with the CRK signal. The TDC signal is a signal indicating that the piston (not shown) of each cylinder is at a predetermined crank angle position slightly before the TDC position of the intake stroke, and one pulse is output for each predetermined crank angle. Is done.

On the other hand, on the downstream side of the exhaust manifold 7a of the exhaust pipe 7 (exhaust passage), first and second catalytic devices 8a and 8b (catalyst) are provided in order from the upstream side with a gap therebetween. Each catalyst device 8 is a combination of a NOx catalyst and a three-way catalyst, and this NOx catalyst is not shown, but an iridium catalyst (a silicon carbide whisker powder carrying iridium and a fired body of silica) is formed on a honeycomb structure. The surface of the material is coated, and a perovskite double oxide (LaCoO ₃ powder and silica fired body) is further coated thereon. The catalyst device 8 purifies NOx in the exhaust gas during the lean burn operation by the oxidation / reduction action of the NOx catalyst, and CO 2 in the exhaust gas during the operation other than the lean burn operation by the oxidation / reduction action of the three-way catalyst. , Purifies HC and NOx. The catalyst device 8 is not limited to a combination of a NOx catalyst and a three-way catalyst, and may be any device that can purify CO, HC, and NOx in the exhaust gas. For example, the catalyst device 8 may be composed of a non-metallic catalyst such as a perovskite catalyst and / or a metal catalyst such as a three-way catalyst.

  An oxygen concentration sensor (hereinafter referred to as “O2 sensor”) 15 is attached between the first and second catalytic devices 8a and 8b. The O2 sensor 15 (downstream air-fuel ratio sensor) is composed of zirconia, a platinum electrode, and the like, and sends an output Vout based on the oxygen concentration in the exhaust gas on the downstream side of the first catalyst device 8a to the ECU 2. The output Vout (control target output) of the O2 sensor 15 is a high level voltage value (for example, 0.8 V) when the air-fuel mixture richer than the stoichiometric air-fuel ratio burns, and when the air-fuel mixture is lean, the output Vout is low. When the air-fuel mixture is near the stoichiometric air-fuel ratio, a predetermined target value Vop (eg, 0.6 V) between the high level and the low level is obtained (see FIG. 2). .

  In addition, a LAF sensor 14 (upstream air-fuel ratio sensor) is attached in the vicinity of the aggregate portion of the exhaust manifold 7a upstream of the first catalyst device 8a. The LAF sensor 14 is configured by combining a sensor similar to the O2 sensor 15 and a detection circuit such as a linearizer. The LAF sensor 14 controls the oxygen concentration in the exhaust gas in a wide range of air-fuel ratios from the rich region to the lean region. The output KACT detected linearly and proportional to the oxygen concentration is sent to the ECU 2. This output KACT is expressed as an equivalent ratio proportional to the reciprocal of the air-fuel ratio.

  Next, the relationship between the exhaust gas purification rate of the first catalyst device 8a and the output Vout (voltage value) of the O2 sensor 15 will be described with reference to FIG. This figure shows the output KACT of the LAF sensor 14, that is, the mixture supplied to the engine 3 when the first catalyst device 8 a is in a deteriorated state in which the purifying capability is lowered due to long-term use and in an undegraded state with a high purifying capability. An example of the results of measuring the HC and NOx purification rates of the two first catalyst devices 8a and the output Vout of the O2 sensor 15 when the air air-fuel ratio changes near the stoichiometric air-fuel ratio is shown. In the figure, all the data indicated by a broken line is a measurement result when the first catalyst device 8a is in an undegraded state, and all the data indicated by a solid line is a measurement result when the first catalyst device 8a is in a deteriorated state. It is. Further, the larger the output KACT of the LAF sensor 14, the richer the air-fuel ratio of the air-fuel mixture.

  As shown in the figure, when the first catalytic device 8a is deteriorated, the exhaust gas purification ability is lower than that in the non-deteriorated state, so that the output KACT of the LAF sensor 14 is reduced. At the leaner value KACT1, the output Vout of the O2 sensor 15 crosses the target value Vop. On the other hand, the first catalyst device 8a has the characteristic of purifying HC and NOx most efficiently when the output Vout of the O2 sensor 15 is at the target value Vop regardless of the deterioration / non-deterioration state. Therefore, it can be seen that the exhaust gas can be purified most efficiently by the first catalyst device 8a by controlling the air-fuel ratio of the air-fuel mixture so that the output Vout of the O2 sensor 15 becomes the target value Vop. For this reason, in the air-fuel ratio control described later, the target air-fuel ratio KCMD is controlled so that the output Vout of the O2 sensor 15 converges to the target value Vop.

  Further, an accelerator opening sensor 16, an atmospheric pressure sensor 17, an intake air temperature sensor 18, a vehicle speed sensor 19 and the like are connected to the ECU 2. The accelerator opening sensor 16 detects a depression amount (hereinafter referred to as “accelerator opening”) AP of an accelerator pedal (not shown) of the vehicle, and outputs a detection signal to the ECU 2. The atmospheric pressure sensor 17, the intake air temperature sensor 18, and the vehicle speed sensor 19 detect the atmospheric pressure PA, the intake air temperature TA, and the vehicle speed VP, respectively, and output detection signals to the ECU 2.

Next, ECU 2 (predicted value calculation means, control input calculation means, gain parameter detection means, gain setting means, air-fuel ratio calculation means, operating state detection means, intermediate value calculation means, target air-fuel ratio calculation means, multiplication means, correction coefficient setting hand stage) will be described.

  The ECU 2 includes a microcomputer including an I / O interface, a CPU, a RAM, and a ROM. The ECU 2 determines the operating state of the engine 3 according to the outputs of the various sensors 10 to 19 described above, and the ROM. The target air-fuel ratio KCMD (control input) is calculated by executing an adaptive air-fuel ratio control process or map search process, which will be described later, according to a control program stored in advance in FIG. Further, as will be described later, the final fuel injection amount TOUT of the injector 6 is calculated for each cylinder based on the target air-fuel ratio KCMD, and the injector 6 is driven with a drive signal based on the calculated final fuel injection amount TOUT. By doing so, the air-fuel ratio of the air-fuel mixture is controlled.

  As shown in FIG. 3, the control device 1 includes an ADSM controller 20 and a PRISM controller 21 that calculate a target air-fuel ratio KCMD, and both the controllers 20 and 21 are specifically configured by an ECU 2. Yes.

  Hereinafter, the ADSM controller 20 (control input calculation means) will be described. The ADSM controller 20 is used for converging the output Vout of the O2 sensor 15 to the target value Vop by a control algorithm for adaptive prediction delta sigma modulation control (hereinafter referred to as “ADSM”) processing described below. The target air-fuel ratio KCMD is calculated, and includes a state predictor 22, an onboard identifier 23, and a DSM controller 24. A specific program for this ADSM processing will be described later.

  First, the state predictor 22 (predicted value calculation means) will be described. The state predictor 22 predicts (calculates) a predicted value PREVO2 of the output deviation VO2 by a prediction algorithm described below. In the present embodiment, the control input to the control target is the target air-fuel ratio KCMD of the air-fuel mixture, the output of the control target is the output Vout of the O2 sensor 15, and the first catalytic device 8 a is taken from the intake system of the engine 3 including the injector 6. The system up to the O2 sensor 15 on the downstream side of the first catalytic device 8a of the exhaust system including the exhaust gas is considered as a control target, and this control target is an ARX that is a discrete time system model as shown in the following equation (1). Model as a model (auto-regressive model with exogeneous input).

VO2 (k) = a1 · VO2 (k-1) + a2 · VO2 (k-2) + b1 · DKCMD (k-dt) ...... (1)
Here, VO2 represents an output deviation which is a deviation (Vout−Vop) between the output Vout of the O2 sensor 15 and the target value Vop described above, and DKCMD is a difference between the target air-fuel ratio KCMD (= φop) and the reference value FLAFBASE. An air-fuel ratio deviation which is a deviation (KCMD-FLAFBASE) is represented, and a symbol k represents an order of sampling cycles of each data. This reference value FLAFBASE is set to a predetermined constant value. Further, a1, a2 and b1 represent model parameters, which are sequentially identified by the onboard identifier 23 as will be described later.

Furthermore, dt in the above equation (1) represents the estimated time from when the air-fuel mixture of the target air-fuel ratio KCMD is supplied to the intake system by the injector 6 until it is reflected in the output Vout of the O2 sensor 15. It is defined as equation (2).
dt = d + d ′ + dd (2)
Here, d is the dead time of the exhaust system from the LAF sensor 14 to the O2 sensor 15, d 'is the dead time of the air-fuel ratio operation system from the injector 6 to the LAF sensor 14, and dd is the empty time between the exhaust system and the exhaust system. The phase delay time with respect to the fuel ratio operation system is shown respectively (in addition, in the control program for adaptive air-fuel ratio control processing described later, processing for calculating the target air-fuel ratio KCMD is performed by switching between ADSM processing and PRISM processing. The phase delay time dd = 0 is set).

  As described above, the reason why the control target model is composed of the time series data of the output deviation VO2 and the air-fuel ratio deviation DKCMD is as follows. In other words, in general, in the control target model, the deviation between the input / output of the control target and the predetermined value is defined as a variable representing the input / output rather than the absolute value of the input / output defined as a variable. It is known that the parameters can be more accurately identified or defined so that the dynamic characteristics of the controlled object model can be adapted to the actual dynamic characteristics of the controlled object. Therefore, as in the control device 1 of the present embodiment, the control target model is constituted by the time series data of the output deviation VO2 and the air-fuel ratio deviation DKCMD, whereby the absolute value of the output Vout of the O2 sensor 15 and the target air-fuel ratio KCMD. Compared with the case where is used as a variable, the adaptability of the dynamic characteristics of the controlled object model to the actual dynamic characteristics of the controlled object can be improved, and thereby the calculation accuracy of the predicted value PREVO2 can be improved.

The predicted value PREVO2 is a value obtained by predicting the output deviation VO2 (k + dt) after the predicted time dt has elapsed since the mixture of the target air-fuel ratio KCMD is supplied to the intake system, and is based on the above formula (1). When the calculation formula of the predicted value PREVO2 is derived, the following formula (3) is obtained.
PREVO2 (k) ≒ VO2 (k + dt)
= A1 · VO2 (k + dt-1) + a2 · VO2 (k + dt-2) + b1 · DKCMD (k) (3)

  In this equation (3), it is necessary to calculate VO2 (k + dt−1) and VO2 (k + dt−2) corresponding to future values of the output deviation VO2 (k), and it is difficult to actually program them. Therefore, the matrices A and B are defined as the equations (4) and (5) shown in FIG. 4 using the model parameters a1, a2 and b1, and the recurrence equation of the above equation (3) is used repeatedly. By transforming the above equation (3), equation (6) shown in FIG. 4 is obtained. When this equation (6) is used as a calculation algorithm for the prediction algorithm, that is, the prediction value PREVO2, the prediction value PREVO2 is calculated from the output deviation VO2 and the air-fuel ratio deviation DKCMD.

  Next, when the LAF output deviation DKACT is defined as a deviation (KACT−FLAFBASE) between the output KACT (= φin) of the LAF sensor 14 and the reference value FLAFBASE, the relationship of DKACT (k) = DKCMD (k−d ′) is obtained. Therefore, when this relationship is applied to equation (6) in FIG. 4, equation (7) shown in FIG. 4 is obtained.

  By using the predicted value PREVO2 calculated by the above formula (6) or formula (7) and calculating the target air-fuel ratio KCMD as will be described later, the response delay and the dead time between the input and output of the controlled object are appropriately set. The target air-fuel ratio KCMD can be calculated while compensating. In particular, when the above equation (7) is used as the prediction algorithm, the predicted value PREVO2 is calculated from the output deviation VO2, the LAF output deviation DKACT, and the air-fuel ratio deviation DKCMD, and thus is actually supplied to the first catalyst device 8a. As a value reflecting the state of the air-fuel ratio of the exhaust gas, the predicted value PREVO2 can be calculated, and the calculation accuracy, that is, the prediction accuracy can be improved as compared with the case where the above equation (6) is used. Further, in the case of using Expression (7), when d ′ ≦ 1 can be considered, the predicted value PREVO2 can be calculated from only the output deviation VO2 and the LAF output deviation DKACT without using the air-fuel ratio deviation DKCMD. In the present embodiment, since the LAF sensor 14 is provided in the engine 3, the above equation (7) is adopted as a prediction algorithm.

  The above-described control target model of the equation (1) is defined as a model having the output deviation VO2 and the LAF output deviation DKACT as variables by applying the relationship of DKACT (k) = DKCMD (k−d ′). It is also possible.

Next, the on-board identifier 23 (identification means, identification error calculation means, filtering means, dead time setting means, restriction range setting means, weight parameter setting means, parameter determination means) will be described. The on-board identifier 23 identifies (calculates) the model parameters a1, a2, and b1 of the above-described equation (1) by a sequential identification algorithm described below. Specifically, the model parameter vector θ (k) is calculated by the equations (8) and (9) shown in FIG. In equation (8) in the figure, KP (k) is a vector of gain coefficients, and ide_f (k) is an identification error filter value. In Equation (9), θ (k) ^T represents a transposed matrix of θ (k), and a1 ′ (k), a2 ′ (k), and b1 ′ (k) are before the limit processing described later is performed. Represents model parameters. In the following description, the expression “vector” is omitted as appropriate.

  The identification error filter value ide_f (k) of the above equation (8) is moved to the identification error ide (k) calculated by the equations (11) to (13) shown in FIG. This is a value subjected to average filtering processing. N in the equation (10) in FIG. 5 represents the filter order (an integer of 1 or more) of the moving average filtering process, and VO2HAT (k) in the equation (12) represents the identification value of the output deviation VO2. .

  The reason for using this identification error filter value ide_f (k) is as follows. That is, the control target of the present embodiment uses the target air-fuel ratio KCMD as a control input and the output Vout of the O2 sensor 15 as the control target output, and has a low-pass characteristic as its frequency characteristic. In the controlled object having such a low-pass characteristic, the high-frequency characteristic of the controlled object is emphasized due to the identification algorithm of the on-board identifier 23, specifically, the frequency weight characteristic of the weighted least square algorithm described later. Since the model parameter is identified in the state, the gain characteristic of the controlled object model tends to be lower than the actual gain characteristic of the controlled object. As a result, when the ADSM process or the PRISM process is executed by the control device 1, the control system may be diverged and unstable due to the over gain state.

  Therefore, in this embodiment, in order to appropriately correct the frequency weighting characteristic of the weighted least squares algorithm and to match the gain characteristic of the controlled object model with the actual gain characteristic of the controlled object, the identification error ide (k ) Is used as the identification error filter value ide_f (k) subjected to the moving average filtering process, and the filter order n of the moving average filtering process is set according to the exhaust gas volume AB_SV, as will be described later.

  Further, the vector KP (k) of the gain coefficient in the above-described equation (8) in FIG. 5 is calculated by the equation (14) in FIG. P (k) in this equation (14) is a cubic square matrix defined by equation (15) in FIG.

In the identification algorithm as described above, one of the following four identification algorithms is selected according to the setting of the weight parameters λ1 and λ2 of Expression (15).
That is,
λ1 = 1, λ2 = 0; fixed gain algorithm λ1 = 1, λ2 = 1; least squares algorithm
λ1 = 1, λ2 = λ; gradually decreasing gain algorithm λ1 = λ, λ2 = 1; weighted least squares algorithm where λ is a predetermined value set to 0 <λ <1.

  In the present embodiment, a weighted least square algorithm of these four identification algorithms is employed. This is because, by setting the value of the weight parameter λ1 according to the operating state of the engine 3, specifically the exhaust gas volume AB_SV, the identification accuracy and the convergence speed of the model parameter to the optimum value can be set appropriately. by. For example, in a low load operation state, by setting the value of the weight parameter λ1 close to the value 1 accordingly, that is, by setting the algorithm close to the least squares algorithm, it is possible to ensure good identification accuracy. At the same time, the model parameter can be quickly converged to the optimum value by setting the value of the weight parameter λ1 to a smaller value in the high load operation state than in the low load operation state. As described above, by setting the value of the weight parameter λ1 according to the exhaust gas volume AB_SV, it is possible to appropriately set the identification accuracy and the convergence speed of the model parameter to the optimum value. Post exhaust gas characteristics can be improved.

  In the identification algorithms of the above equations (8) to (15), when the above-described relationship of DKACT (k) = DKCMD (k−d ′) is applied, the identification algorithms of equations (16) to (23) shown in FIG. Is obtained. In the present embodiment, since the LAF sensor 14 is provided in the engine 3, these equations (16) to (23) are used. When these formulas (16) to (23) are used, the model parameter is identified as a value more reflecting the state of the air-fuel ratio of the exhaust gas actually supplied to the first catalyst device 8a for the reason described above. Thus, the identification accuracy of the model parameters can be improved as compared with the case where the identification algorithms of the above formulas (8) to (15) are used.

  In addition, the on-board identifier 23 performs a limit process described later on the model parameters a1 ′ (k), a2 ′ (k), and b1 ′ (k) calculated by the above identification algorithm, so that the model parameters are obtained. a1 (k), a2 (k) and b1 (k) are calculated. Further, in the state predictor 22 described above, the predicted value PREVO2 is calculated based on the model parameters a1 (k), a2 (k), and b1 (k) after the limit processing is performed.

  Next, the DSM controller 24 will be described. The DSM controller 24 generates (calculates) a control input φop (k) (= target air-fuel ratio KCMD) based on the predicted value PREVO2 calculated by the state predictor 22 by a control algorithm applying a ΔΣ modulation algorithm. By inputting this to the control target, the output Vout of the O2 sensor 15 as the control target output is controlled to converge to the target value Vop.

First, a general ΔΣ modulation algorithm will be described. FIG. 7 shows a configuration of a control system in which the control object 27 is controlled by the controller 26 to which the ΔΣ modulation algorithm is applied. As shown in the figure, in this controller 26, a difference signal δ (k) is obtained as a deviation between the reference signal r (k) and the DSM signal u (k−1) delayed by the delay element 26b by the differencer 26a. Generated. Next, an integrator 26c, a deviation integral value σ _d (k) is generated as a signal of the sum of the deviation signal [delta] (k) and an integrated deviation value delayed by a delay element 26d σ _d (k-1) The Next, the quantizer 26e (sign function) generates the DSM signal u (k) as a signal obtained by encoding the deviation integral value σ _d (k). Then, when the DSM signal u (k) generated as described above is input to the control target 27, the output signal y (k) is output from the control target 27.

The above ΔΣ modulation algorithm is expressed by the following mathematical formulas (24) to (26).
δ (k) = r (k) −u (k−1) (24)
σ _d (k) = σ _d (k−1) + δ (k) (25)
u (k) = sgn (σ _d (k)) (26)
However, the value of the sign function sgn (σ _d (k)) _{is, σ d (k) ≧ 0} sgn (σ d (k)) = 1 becomes when _{the, σ d (k) <0} sgn (σ d when the (k)) = − 1 (Note that when σ _d (k) = 0, sgn (σ _d (k)) = 0 may be set).

  FIG. 8 shows a control simulation result of the above control system. As shown in the figure, when a sinusoidal reference signal r (k) is input to the control system, a DSM signal u (k) is generated as a rectangular wave signal, and this is input to the control object 27. An output signal y (k) having the same waveform as that of the reference signal r (k) but having the same frequency and different frequency is output from the control object 27. Thus, the characteristic of the ΔΣ modulation algorithm is that when the DSM signal u (k) generated from the reference signal r (k) is input to the control object 27, the output y (k) of the control object 27 is the reference signal. The point is that the DSM signal u (k) can be generated as a value that has the same waveform with different amplitude and the same frequency with respect to r (k). In other words, the DSM signal u (k) can be generated (calculated) as a value such that the reference signal r (k) is reproduced in the actual output y (k) of the control object 27.

The DSM controller 24 calculates the control input φop (k) for converging the output Vout of the O2 sensor 15 to the target value Vop by using such characteristics of the ΔΣ modulation algorithm. The principle will be described. For example, as shown by a one-dot chain line in FIG. 9, when the output deviation VO2 fluctuates with respect to the value 0 (that is, the output Vout of the O2 sensor 15 fluctuates with respect to the target value Vop). In order to make the output deviation VO2 converge to the value 0 (that is, to make the output Vout converge to the target value Vop), the output deviation VO2 ^* having an antiphase waveform that cancels the output deviation VO2 shown in FIG. The control input φop (k) may be generated so as to occur.

However, as described above, in the control target of the present embodiment, the estimated time from when the target air-fuel ratio KCMD as the control input φop (k) is input to the control target and reflected in the output Vout of the O2 sensor 15 is estimated. Since a time delay of dt occurs, the output deviation VO2 ^# when the control input φop (k) is calculated based on the current output deviation VO2 is the output deviation VO2 ^* as shown by the solid line in FIG. On the other hand, there is a delay, which causes a shift in control timing. Therefore, in order to compensate for this, the DSM controller 24 in the ADSM controller 20 of the present embodiment uses the predicted value PREVO2 of the output deviation VO2 so that the control input φop (k) does not cause a shift in control timing. The signal is generated as a signal that causes an output deviation (an output deviation similar to the output deviation VO2 ^{* of the} antiphase waveform) that cancels the current output deviation VO2.

More specifically, in the DSM controller 24, as shown in FIG. 10, the inverting amplifier 24a, the reference signal r (k), the value -1, the gain G _d and the predicted value of the reference signal PREVO2 (k) of Generated as signals multiplied by each other. Next, the differencer 24b generates a deviation signal δ (k) as a deviation between the reference signal r (k) and the DSM signal u ″ (k−1) delayed by the delay element 24c.

Then, an integrator 24d, integrated deviation value σ _d (k) is generated as a signal of the sum of the deviation signal [delta] (k) and an integrated deviation value delayed by a delay element _{24e σ d (k-1)} , Next, the quantizer 24f (sign function) generates the DSM signal u ″ (k) as a value obtained by encoding the deviation integral value σ _d (k). Then, the amplified DSM signal u (k) is generated as a value obtained by amplifying the DSM signal u ″ (k) with a predetermined gain F _d by the amplifier 24g, and then this amplified DSM signal u ( The control input φop (k) is generated as a value obtained by adding k) to a predetermined reference value FLAFBASE.

The control algorithm of the above DSM controller 24 is expressed by the following equations (27) to (32).
r (k) = − 1 · G _d · PREVO2 (k) (27)
δ (k) = r (k) −u ″ (k−1) (28)
σ _d (k) = σ _d (k−1) + δ (k) (29)
u ″ (k) = sgn (σ _d (k)) (30)
u (k) = F _d · u ″ (k) (31)
φop (k) = FLAFBASE + u (k) (32)
Here, G _d and F _d represent gains. The value of the sign function sgn (σ _d (k)) _{is, σ d (k) ≧ 0} sgn (σ d (k)) = 1 becomes when _{the, σ d (k) <0} sgn (σ d when the (k)) = − 1 (Note that when σ _d (k) = 0, sgn (σ _d (k)) = 0 may be set).

In the DSM controller 24, as described above, the control input φop (k) cancels the output deviation VO2 without causing a shift in control timing, as described above, by the control algorithm shown in the above equations (27) to (32). Is calculated as a value that causes a large output deviation VO2 ^* . That is, the control input φop (k) is calculated as a value that can converge the output Vout of the O2 sensor 15 to the target value Vop. Further, since the control input φop (k) is calculated as a value obtained by adding the amplified DSM signal u (k) to the predetermined reference value FLAFBASE, only the value that reverses the positive / negative of the control input φop (k) around the value 0 is obtained. Instead, it can be calculated as a value that repeatedly increases and decreases around the reference value FLAFBASE. Thereby, the degree of freedom of control can be increased as compared with a normal ΔΣ modulation algorithm.

  Next, the PRISM controller 21 will be described. The PRISM controller 21 sets a target air-fuel ratio KCMD for converging the output Vout of the O2 sensor 15 to the target value Vop by a control algorithm of an on-board identification type sliding mode control process (hereinafter referred to as “PRISM process”) described below. The calculation is made up of a state predictor 22, an on-board identifier 23, and a sliding mode controller (hereinafter referred to as “SLD controller”) 25. A specific program for this PRISM processing will be described later.

Since the state predictor 22 and the onboard identifier 23 in the PRISM controller 21 have already been described, only the SLD controller 25 will be described here. The SLD controller 25 performs sliding mode control based on a sliding mode control algorithm, and a general sliding mode control algorithm will be described below. In this sliding mode control algorithm, since the discrete time system model of the above-described equation (1) is used as the control target model, the switching function σ is linear in the time series data of the output deviation VO2, as shown in the following equation (33). Set as a function.
σ (k) = S1 · VO2 (k) + S2 · VO2 (k−1) (33)
Here, S1 and S2 are predetermined coefficients set so that the relationship of -1 <(S2 / S1) <1 is established.

  In general, in the sliding mode control algorithm, when the switching function σ is composed of two state variables (in this embodiment, time series data of the output deviation VO2), the phase space composed of the two state variables respectively Since the vertical axis and the horizontal axis are two-dimensional phase planes, the combination of the two state variable values satisfying σ = 0 is placed on a straight line called a switching straight line. Therefore, by appropriately determining the control input to the controlled object so that the combination of the two state variables converges on the switching straight line, both of the two state variables become equilibrium points where the value becomes zero. It can be converged (sliding). Furthermore, in the sliding mode control algorithm, it is possible to specify the dynamic characteristics of the state variable, more specifically the convergence behavior and the convergence speed, by setting the switching function σ. For example, as in the present embodiment, when the switching function σ is composed of two state variables, when the slope of the switching line is brought close to the value 1, the convergence speed of the state variable becomes slow, while the value 0 is reached. The closer it is, the faster the convergence speed.

  In the present embodiment, as shown in the equation (33), the switching function σ is based on two time-series data of the output deviation VO2, that is, the current value VO2 (k) and the previous value VO2 (k−1) of the output deviation VO2. If the control input to the controlled object, that is, the target air-fuel ratio KCMD is set so that the combination of the current value VO2 (k) and the previous value VO2 (k-1) converges on the switching line, Good. Specifically, when the control amount Usl (k) is defined as a value that is the sum of the reference value FLAFBASE and the target air-fuel ratio KCMD, the combination of the current value VO2 (k) and the previous value VO2 (k-1) is switched. The control amount Usl (k) for converging on the straight line is obtained from an equivalent control input Ueq (k), a reaching law input Urch (k), and an reaching control input Urch (k) by an adaptive sliding mode control algorithm as shown in Expression (34) shown in FIG. It is set as the sum of the adaptive law input Uadp (k).

  This equivalent control input Ueq (k) is for constraining the combination of the current value VO2 (k) and the previous value VO2 (k−1) of the output deviation VO2 on the switching straight line. , Is defined as shown in Expression (35) shown in FIG. Further, the reaching law input Urch (k) is generated when the combination of the current value VO2 (k) and the previous value VO2 (k−1) of the output deviation VO2 deviates from the switching straight line due to disturbance or modeling error. These are for convergence on the switching straight line, and are specifically defined as in Expression (36) shown in FIG. In this formula (36), F represents a gain.

  Further, the adaptive law input Uadp (k) is a combination of the current value VO2 (k) and the previous value VO2 (k−1) of the output deviation VO2 while suppressing the influence of the steady-state deviation, modeling error and disturbance of the controlled object. Is reliably converged on the switching hyperplane, and is specifically defined as shown in Expression (37) shown in FIG. In this equation (37), G represents a gain, and ΔT represents a control period.

  In the SLD controller 25 of the PRISM controller 21 of the present embodiment, as described above, the predicted value PREVO2 is used instead of the output deviation VO2, so that by applying the relationship of PREVO2 (k) ≈VO2 (k + dt) The algorithms of equations (33) to (37) are rewritten and used as equations (38) to (42) shown in FIG. In this equation (38), σPRE is a value of a switching function (hereinafter referred to as “predictive switching function”) when the predicted value PREVO2 is used. That is, the SLD controller 25 calculates the target air-fuel ratio KCMD by adding the control amount Usl (k) calculated by the above algorithm to the reference value FLAFBASE.

  Hereinafter, the fuel injection amount calculation process executed by the ECU 2 will be described with reference to FIG. In the following description, the symbol (k) indicating the current value is omitted as appropriate. FIG. 13 shows the main routine of this control process, and this process is interrupted in synchronization with the input of the TDC signal. In this process, the fuel injection amount TOUT is calculated for each cylinder by using a target air-fuel ratio KCMD calculated by an adaptive air-fuel ratio control process or map search process described later.

  First, in step 1 (abbreviated as “S1” in the figure, the same applies hereinafter), the outputs of the various sensors 10 to 19 described above are read, and the read data is stored in the RAM.

  Next, the process proceeds to step 2, and the basic fuel injection amount Tim is calculated. In this process, the basic fuel injection amount Tim is calculated by searching a map (not shown) according to the engine speed NE and the intake pipe absolute pressure PBA.

  Next, the process proceeds to step 3 where a total correction coefficient KTOTAL is calculated. This total correction coefficient KTOTAL is obtained by searching various tables and maps according to various operation parameters (for example, intake air temperature TA, atmospheric pressure PA, engine water temperature TW, accelerator pedal opening AP, etc.). Is calculated by multiplying these various correction coefficients with each other.

Next, the process proceeds to step 4 where an adaptive control flag F_PRISMON setting process is executed. Although details of this process are not shown, specifically, when all of the following conditions (a) to (f) are satisfied, the target air-fuel ratio KCMD calculated in the adaptive air-fuel ratio control process is used. Assuming that the condition is satisfied, the adaptive control flag F_PRISMON is set to “1” to represent it. On the other hand, when at least one of the conditions (a) to (f) is not satisfied, the adaptive control flag F_PRISMON is set to “0”.
(A) Both the LAF sensor 14 and the O2 sensor 15 are activated.
(B) The engine 3 is not in lean burn operation.
(C) The throttle valve 5 is not fully opened.
(D) The ignition timing is not retarded.
(E) The fuel cut operation is not in progress.
(F) Both the engine speed NE and the intake pipe absolute pressure PBA are values within a predetermined range.

  Next, the process proceeds to step 5 where it is determined whether or not the adaptive control flag F_PRISMON set in step 4 is “1”. When the determination result is YES, the process proceeds to Step 6 in which the target air-fuel ratio KCMD is set to an adaptive target air-fuel ratio KCMDLSD calculated by an adaptive air-fuel ratio control process described later.

  On the other hand, when the determination result of step 5 is NO, the process proceeds to step 7 and the target air-fuel ratio KCMD is set to the map value KCMDMAP. The map value KCMDMAP is calculated by searching a map (not shown) according to the engine speed NE and the intake pipe absolute pressure PBA.

  In step 8 following step 6 or 7 described above, the observer feedback correction coefficient #nKLAF is calculated for each cylinder. This observer feedback correction coefficient #nKLAF is for correcting the variation in the actual air-fuel ratio for each cylinder, and specifically, the actual air-fuel ratio for each cylinder is estimated from the output KACT of the LAF sensor 14 by the observer. Then, it is calculated by PID control according to these estimated air-fuel ratios. Note that the symbol #n of the observer feedback correction coefficient #nKLAF represents the cylinder numbers # 1 to # 4, and this also applies to the requested fuel injection amount #nTCYL and the final fuel injection amount #nTOUT described later. It is.

  Next, the process proceeds to step 9 to calculate a feedback correction coefficient KFB. Specifically, the feedback correction coefficient KFB is calculated as follows. That is, the feedback coefficient KLAF is calculated by PID control according to the deviation between the output KACT of the LAF sensor 14 and the target air-fuel ratio KCMD. Further, the feedback correction coefficient KSTR is calculated by a Self Tuning Regulator type adaptive controller (not shown), and this is divided by the target air-fuel ratio KCMD to calculate the feedback correction coefficient kstr. Then, one of these two feedback coefficients KLAF and feedback correction coefficient kstr is set as the feedback correction coefficient KFB according to the operating state of the engine 3.

  Next, the routine proceeds to step 10 where a corrected target air-fuel ratio KCMDM is calculated. The corrected target air-fuel ratio KCMDM is for compensating for the change in charging efficiency due to the change in the air-fuel ratio A / F, and is a table (not shown) according to the target air-fuel ratio KCMD calculated in step 6 or 7 described above. It is calculated by searching.

Next, the process proceeds to step 11, and the basic fuel injection amount Tim, the total correction coefficient KTOTAL, the observer feedback correction coefficient #nKLAF, the feedback correction coefficient KFB, and the corrected target air-fuel ratio KCMDM calculated as described above are used. ) To calculate the required fuel injection amount #nTCYL for each cylinder.
# NTCYL = Tim, KTOTAL, KCMDM, KFB, #nKLAF
...... (43)

  Next, the process proceeds to step 12, and the final fuel injection amount #nTOUT is calculated by adhering the required fuel injection amount #nTCYL. Specifically, the final fuel injection amount #nTOUT is calculated according to the operation state, such as the ratio of the fuel injected from the injector 6 in the current combustion cycle adhering to the inner wall surface of the combustion chamber. It is calculated by correcting the required fuel injection amount #nTCYL based on the calculated ratio.

  Next, the routine proceeds to step 13 where a drive signal based on the final fuel injection amount #nTOUT calculated as described above is output to the injector 6 of the corresponding cylinder, and then this processing is terminated.

  Next, the adaptive air-fuel ratio control process including the ADSM process and the PRISM process will be described with reference to FIGS. 14 and 15. This process is executed at a predetermined cycle (for example, 10 msec). In this process, the target air-fuel ratio KCMD is calculated by the ADSM process, PRISM process, or the process of setting the sliding mode control amount DKCMDSLD to the predetermined value SLDHOLD according to the operating state of the engine 3.

  In this process, first, in step 20, post-F / C determination processing is executed. Although the contents of this process are not shown in the figure, in this process, during the fuel cut operation, the post-F / C determination flag F_AFC is set to “1”, and after the fuel cut operation is completed, a predetermined time X_TM_AFC is set. When the time has elapsed, the post-F / C determination flag F_AFC is set to “0” to indicate that.

  Next, it progresses to step 21 and the start determination process which determines whether the vehicle carrying the engine 3 started based on the vehicle speed VP is performed. As shown in FIG. 16, in this process, first, in step 49, it is determined whether or not an idle operation flag F_IDLE is “1”. The idle operation flag F_IDLE is set to “1” when idling, and to “0” otherwise.

  When the determination result is YES and the idling operation is being performed, the process proceeds to step 50 to determine whether or not the vehicle speed VP is lower than a predetermined vehicle speed VSTART (for example, 1 km / h). If the determination result is YES and the vehicle is stopped, the routine proceeds to step 51, where the timer value TMVOTVST of the down-counting first start determination timer is set to a first predetermined time TVOTVST (for example, 3 msec).

  Next, the routine proceeds to step 52, where the timer value TMVST of the down-count type second start determination timer is set to a second predetermined time TVST (for example, 500 msec) longer than the first predetermined time TVOTVST. Next, in steps 53 and 54, the first and second start flags F_VOTVST and F_VST are both set to “0”, and then the present process is terminated.

  On the other hand, when the determination result of step 49 or 50 is NO, that is, when the vehicle is not idling or when the vehicle starts, the process proceeds to step 55, and whether or not the timer value TMVOTVST of the first start determination timer is greater than 0. Is determined. If the determination result is YES and the first predetermined time TVOTVST has not elapsed after the end of the idle operation or the start of the vehicle, it is determined that the first start mode is in progress, and the process proceeds to step 56 to indicate it. 1 start flag F_VOTVST is set to “1”.

  On the other hand, if the determination result in step 55 is NO and the first predetermined time TVOTVST has elapsed after the end of the idle operation or the start of the vehicle, it is determined that the first start mode has ended, the process proceeds to step 57, and the first start flag F_VOTVST is set to “0”.

  In step 58 following step 56 or 57, it is determined whether or not the timer value TMVST of the second start determination timer is greater than 0. If this determination result is YES and the second predetermined time TVST has not elapsed after the end of the idle operation or the start of the vehicle, it is determined that the second start mode is in progress, and the process proceeds to step 59 to express it. After the 2 start flag F_VST is set to “1”, this process is terminated.

  On the other hand, if the determination result in step 58 is NO and the second predetermined time TVST has elapsed after the end of the idle operation or the start of the vehicle, the second start mode is determined to have ended, The process ends.

  Returning to FIG. 14, in step 22 following step 21, a state variable setting process is executed. Although not shown, in this process, the time series data of the target air-fuel ratio KCMD, the output KACT of the LAF sensor 14 and the output deviation VO2 stored in the RAM are all shifted to the past by one sampling cycle. Thereafter, current values of KCMD, KACT, and VO2 are calculated based on the latest values of the time series data of KCMD, KACT, and VO2, a reference value FLAFBASE, and an adaptive correction term FLAFADP described later.

  Next, the process proceeds to step 23 where execution determination processing for PRISM / ADSM processing is performed. This process determines whether or not the execution condition for the PRISM process or the ADSM process is satisfied, and is specifically executed as shown in the flowchart of FIG.

That is, in steps 60 to 63 in FIG. 17, when any of the following conditions (g) to (j) is satisfied, it is indicated that the PRISM process or the ADSM process is in an operating state. In step 64, the PRISM / ADSM execution flag F_PRISMCAL is set to “1”, and the process is terminated. On the other hand, when at least one of the conditions (g) to (j) is not satisfied, it is determined that the PRISM process or the ADSM process is not in the operating state to be executed, and in order to indicate that, the PRISM / ADSM execution is performed in step 65. After the flag F_PRISMCAL is set to “0”, this process is terminated.
(G) The O2 sensor 15 is activated.
(H) The LAF sensor 14 is activated.
(I) The engine 3 is not in lean burn operation.
(J) The ignition timing is not retarded.

  Returning to FIG. 14, in step 24 following step 23, execution determination processing for the identifier operation is performed. This process is to determine whether or not the parameter identification execution condition by the on-board identifier 23 is satisfied, and is specifically executed as shown in the flowchart of FIG.

  That is, when the determination results of steps 70 and 71 in FIG. 18 are both NO, in other words, when the throttle valve opening θTH is not in the fully open state and the fuel cut operation is not being performed, the parameter identification should be performed. If there is, the process proceeds to step 72, the identification execution flag F_IDCAL is set to “1”, and then this process ends. On the other hand, when the determination result in step 70 or 71 is YES, it is determined that there is no operating state in which parameter identification is to be performed, the process proceeds to step 73, the identification execution flag F_IDCAL is set to “0”, and this process is terminated.

  Returning to FIG. 14, in step 25 following step 24, various parameters (exhaust gas volume AB_SV, etc.) are calculated. The specific contents of this process will be described later.

  Next, the process proceeds to step 26, where it is determined whether or not the PRISM / ADSM execution flag F_PRISMCAL set in step 23 is “1”. If the determination result is YES and the execution condition of the PRISM process or ADSM process is satisfied, the process proceeds to step 27 to determine whether or not the identification execution flag F_IDCAL set in step 24 is “1”. .

  If the determination result is YES and the operation state is to execute parameter identification by the on-board identifier 23, the process proceeds to step 28 to determine whether or not the parameter initialization flag F_IDRSET is “1”. If the determination result is NO and initialization of the model parameters a1, a2, and b1 stored in the RAM is unnecessary, the process proceeds to step 31 described later.

  On the other hand, if the determination result is YES and the model parameters a1, a2, b1 need to be initialized, the process proceeds to step 29, where the model parameters a1, a2, b1 are set to their initial values and then In order to represent it, the process proceeds to step 30, and the parameter initialization flag F_IDRSET is set to “0”.

  In step 31 following step 30 or 28, the computation of the on-board identifier 23 is executed to identify the model parameters a1, a2, and b1, and then the process proceeds to step 32 in FIG. Specific contents of the calculation of the on-board identifier 23 will be described later.

  On the other hand, if the determination result in step 27 is NO and the operation state is not to be executed, parameter determination is skipped and the process proceeds to step 32 in FIG. In step 32 following step 27 or 31, an identification value or a predetermined value is selected as model parameters a1, a2 and b1. Although details of this processing are not shown, specifically, when the identification execution flag F_IDCAL set in step 24 is “1”, the model parameters a1, a2, and b1 are set to the identification values identified in step 31. To do. On the other hand, when the identification execution flag F_IDCAL is “0”, the model parameters a1, a2, and b1 are set to predetermined values.

  Next, the process proceeds to step 33, where the state predictor 22 is operated to calculate a predicted value PREVO2 as will be described later. Thereafter, the process proceeds to step 34, and the control amount Usl is calculated as will be described later.

  Next, the routine proceeds to step 35, where the stability determination of the SLD controller 25 is executed. Although details of this processing are not shown, specifically, it is determined whether or not the sliding mode control by the SLD controller 25 is in a stable state based on the value of the prediction switching function σPRE.

  Next, in steps 36 and 37, as will be described later, the sliding mode control amount DKCMDSLD and the ΔΣ modulation control amount DKCMDDSM are calculated by the SLD controller 25 and the DSM controller 24, respectively.

  Next, the routine proceeds to step 38, where the adaptive target air-fuel ratio KCMDLSLD is calculated using the sliding mode control amount DKCMDDSLD calculated by the SLD controller 25 or the ΔΣ modulation control amount DKCMDDSM calculated by the DSM controller 24, as will be described later. . Thereafter, the process proceeds to step 39, and as will be described later, after the adaptive correction term FLAFADP is calculated, this process is terminated.

  On the other hand, returning to FIG. 14, if the determination result in step 26 is NO and neither of the execution conditions of the PRISM process and ADSM process is satisfied, the process proceeds to step 40 and the parameter initialization flag F_IDRSET is set to “1”. To do. Next, the process proceeds to step 41 in FIG. 15, and the sliding mode control amount DKCMDSLD is set to a predetermined value SLDHOLD. Next, after executing steps 38 and 39 described above, the present process is terminated.

Next, the process for calculating various parameters in step 25 will be described with reference to FIG. In this process, first, in step 80, the exhaust gas volume AB_SV (estimated space velocity) is calculated by the following equation (44).
AB_SV = (NE / 1500) · PBA · X_SVPRA (44)
Here, X_SVPRA is a predetermined coefficient determined based on the engine displacement.

  Next, the routine proceeds to step 81, where the above-described dead time KACT_D (= d ′) of the air-fuel ratio operation system, the dead time CAT_DELAY (= d) of the exhaust system, and the predicted time dt are calculated. Specifically, the dead time KACT_D and CAT_DELAY are calculated by searching the table shown in FIG. 20 according to the exhaust gas volume AB_SV calculated in step 80, and the sum (KACT_D + CAT_DELAY) of these is estimated time. Set as dt. That is, in this control program, the phase delay time dd is set to the value 0.

  In this table, the dead times KACT_D and CAT_DELAY are set to smaller values as the exhaust gas volume AB_SV is larger. This is because the dead time KACT_D, CAT_DELAY is shortened by increasing the exhaust gas flow velocity as the exhaust gas volume AB_SV is increased. As described above, the dead times KACT_D and CAT_DELAY and the predicted time dt are calculated according to the exhaust gas volume, and therefore, based on the predicted value PREVO2 of the output deviation VO2 calculated using these, the adaptive target air-fuel ratio KCMDLSLD described later. By calculating, it is possible to eliminate the control timing shift between the input and output of the control target. In addition, since the model parameters a1, a2, and b1 are identified using the dead time CAT_DELAY, the dynamic characteristics of the controlled object model can be adapted to the actual dynamic characteristics of the controlled object, and thereby the controlled object. The control timing shift between the input and output can be further eliminated.

  Next, proceeding to step 82, the values of the weighting parameters λ1 and λ2 of the identification algorithm are calculated. Specifically, the weight parameter λ2 is set to the value 1, and the weight parameter λ1 is calculated by searching the table shown in FIG. 21 according to the exhaust gas volume AB_SV.

  In this table, the larger the exhaust gas volume AB_SV, the smaller the weight parameter λ1 is set. In other words, the smaller the exhaust gas volume AB_SV is, the larger the weight parameter λ1 is set to a value closer to the value 1. Has been. This is because the model parameter needs to be identified more quickly as the exhaust gas volume AB_SV is larger, in other words, as the engine is operating at a higher load. Therefore, by setting the weight parameter λ1 smaller, the model parameter can be optimized. This is to increase the convergence speed to the value. In addition to this, the smaller the exhaust gas volume AB_SV, that is, the lower the load operating state, the more easily the air-fuel ratio fluctuates and the post-catalyst exhaust gas characteristics tend to become unstable. This is because accuracy needs to be ensured, so that the identification accuracy of the model parameter is further increased by bringing the weight parameter λ1 closer to the value 1 (closer to the least square algorithm).

  Next, the routine proceeds to step 83 where the lower limit value X_IDA2L for limiting the values of the model parameters a1 and a2, and the lower limit value X_IDB1L and the upper limit value X_IDB1H for limiting the value of the model parameter b1 are set to the exhaust gas volume AB_SV. Accordingly, the calculation is performed by searching the table shown in FIG.

  In this table, the lower limit value X_IDA2L is set to a larger value as the exhaust gas volume AB_SV is larger. This is because the combination of the model parameters a1 and a2 at which the control system becomes stable changes as the dead time increases or decreases according to the change in the exhaust gas volume AB_SV. Further, the lower limit value X_IDB1L and the upper limit value X_IDB1H are also set to larger values as the exhaust gas volume AB_SV is larger. This is because the larger the exhaust gas volume AB_SV is, the degree of the influence of the pre-catalyst air-fuel ratio (the air-fuel ratio of the exhaust gas upstream of the first catalyst device 8a) on the output Vout of the O2 sensor 15, that is, the gain to be controlled. By becoming larger.

  Next, the process proceeds to step 84, and after calculating the filter order n of the moving average filtering process, this process ends. In this process, the filter order n is calculated by searching the table shown in FIG. 23 according to the exhaust gas volume AB_SV.

  In this table, the filter order n is set to a smaller value as the exhaust gas volume AB_SV is larger. This is due to the following reason. That is, as described above, when the exhaust gas volume AB_SV changes, the frequency characteristic, particularly the gain characteristic, of the controlled object changes, so that the gain characteristic of the controlled object model matches the actual gain characteristic of the controlled object. It is necessary to appropriately correct the frequency weighting characteristic of the weighted least squares algorithm according to the exhaust gas volume AB_SV. Therefore, by setting the filter order n of the moving average filtering process according to the exhaust gas volume AB_SV as in the above table, a constant identification weight can be secured in the identification algorithm regardless of the change in the exhaust gas volume AB_SV. The gain characteristics of the controlled object model and the controlled object can be made to coincide with each other, thereby improving the identification accuracy.

Next, the calculation process of the on-board identifier 23 in the step 31 will be described with reference to FIG. As shown in the figure, in this process, first, in step 90, the gain coefficient KP (k) is calculated from the above-described equation (22). Next, the process proceeds to step 91, where an identification value VO2HAT (k) of the output deviation VO2 is calculated from the above-described equation (20).

  Next, the routine proceeds to step 92, where the identification error filter value ide_f (k) is calculated from the aforementioned equations (18) and (19). Next, the process proceeds to step 93, and after calculating the model parameter vector θ (k) from the above-described equation (16), the process proceeds to step 94, and the stabilization process of the model parameter vector θ (k) is executed. This process will be described later.

  Next, the routine proceeds to step 95, where the next value P (k + 1) of the square matrix P (k) is calculated from the aforementioned equation (23). This next value P (k + 1) is used as the value of the square matrix P (k) in the next calculation in the loop.

  The process for stabilizing the model parameter vector θ (k) in step 94 will be described below with reference to FIG. As shown in the figure, first, in step 100, the three flags F_A1STAB, F_A2STAB, and F_B1STAB are all set to “0”.

  Next, the process proceeds to step 101, and a limit process of a1 ′ & a2 ′ is executed as will be described later. Next, at step 102, as described later, after executing the limit process of b1 ′, this process is terminated.

  Hereinafter, the a1 ′ & a2 ′ limit process in step 101 will be described with reference to FIG. As shown in the figure, first, at step 110, it is determined whether or not the model parameter identification value a2 ′ calculated at step 93 is equal to or greater than the lower limit value X_IDA2L calculated at step 83 of FIG. . When the determination result is NO, the process proceeds to step 111, and in order to stabilize the control system, the model parameter a2 is set to the lower limit value X_IDA2L, and at the same time, a flag is used to indicate that the model parameter a2 has been stabilized. F_A2STAB is set to “1”. On the other hand, when the determination result is YES and a2 ′ ≧ X_IDA2L, the process proceeds to step 112, and the model parameter a2 is set to the identification value a2 ′.

  In step 113 following these steps 111 or 112, is the identification value a1 ′ of the model parameter calculated in step 93 greater than or equal to a predetermined lower limit value X_IDA1L (for example, a constant value greater than or equal to −2 and less than 0)? Determine whether or not. When the determination result is NO, the process proceeds to step 114, and in order to stabilize the control system, the model parameter a1 is set to the lower limit value X_IDA1L, and at the same time, a flag is used to indicate that the model parameter a1 has been stabilized. F_A1STAB is set to “1”.

  On the other hand, when the determination result of step 113 is YES, the process proceeds to step 115 to determine whether or not the identification value a1 ′ is equal to or less than a predetermined upper limit value X_IDA1H (for example, value 2). When the determination result is YES and X_IDA1L ≦ a1 ′ ≦ X_IDA1H, the process proceeds to step 116, and the model parameter a1 is set to the identification value a1 ′. On the other hand, when the determination result is NO and X_IDA1H <a1 ′, the process proceeds to step 117 to set the model parameter a1 to the upper limit value X_IDA1H, and at the same time, to indicate that the model parameter a1 has been stabilized, the flag F_A1STAB Is set to “1”.

  In step 118 following these steps 114, 116 or 117, the sum (| a1 | + a2) of the absolute value of the model parameter a1 calculated as described above and the model parameter a2 (| a1 | + a2) is a predetermined determination value X_A2STAB (for example, a value). 0.9) It is determined whether or not it is less than or equal to. When the determination result is YES, it is determined that the combination of the model parameters a1 and a2 is within a range in which the stability of the control system can be ensured (a regulation range indicated by hatching in FIG. 27), and the present process is terminated.

  On the other hand, when the determination result in step 118 is NO, the process proceeds to step 119, where it is determined whether or not the model parameter a1 is equal to or less than a value obtained by subtracting the lower limit value X_IDA2L from the determination value X_A2STAB (X_A2STAB-X_IDA2L). When the determination result is YES, the process proceeds to step 120 where the model parameter a2 is set to a value obtained by subtracting the absolute value of the model parameter a1 from the determination value X_A2STAB (X_A2STAB- | a1 |) and at the same time, the model parameter a2 is stabilized. Is set to “1” to end this processing.

  On the other hand, if the determination result in step 119 is NO and a1> (X_A2STAB-X_IDA2L), the process proceeds to step 121, and the model parameter a1 is subtracted from the determination value X_A2STAB with the lower limit value X_IDA2L in order to stabilize the control system. Value (X_A2STAB-X_IDA2L), and the model parameter a2 is set to the lower limit value X_IDA2L. At the same time, flags F_A1STAB and F_A2STAB are both set to “1” to indicate that the model parameters a1 and a2 have been stabilized. Thereafter, this process is terminated.

  As described above, in the sequential identification algorithm, when the input / output of the controlled object is in a steady state, the so-called drift phenomenon in which the absolute value of the identified model parameter increases due to insufficient self-excitation conditions occurs. When it occurs easily, the control system may become unstable or vibrate. The stability limit also changes according to the operating state of the engine 3. For example, in the low-load operation state, the exhaust gas volume AB_SV is reduced, so that the response delay or dead time of the exhaust gas with respect to the supplied air-fuel mixture is increased, thereby causing the output Vout of the O2 sensor 15 to vibrate. It is easy to become.

  On the other hand, in the limit processing of a1 ′ & a2 ′ described above, the combination of the model parameters a1 and a2 is set so as to fall within the regulation range indicated by hatching in FIG. 27, and this regulation range is determined. Since the lower limit value X_IDA2L to be set is set according to the exhaust gas volume AB_SV, this restriction range is appropriately changed to reflect the change in the stability limit accompanying the change in the operating state of the engine 3, that is, the change in the dynamic characteristics of the controlled object. The model parameters a1 and a2 that can be set as the stability limit range and are regulated so as to be within such a regulation range can avoid the occurrence of the drift phenomenon and ensure the stability of the control system. can do. In addition to this, by setting the combination of the model parameters a1 and a2 as a value within the above-mentioned regulation range that can ensure the stability of the control system, the control when the model parameter a1 and the model parameter a2 are regulated independently Generation of unstable system conditions can be avoided. As described above, the stability of the control system can be improved, and the post-catalyst exhaust gas characteristics can be improved.

  Next, referring to FIG. 28, the limit process of b1 ′ in step 102 will be described. As shown in the figure, in this process, in step 130, whether or not the model parameter identification value b1 ′ calculated in step 93 is equal to or greater than the lower limit value X_IDB1L calculated in step 83 of FIG. Is determined.

  If the determination result is YES and b1 ′ ≧ X_IDB1L, the process proceeds to step 1311 to determine whether or not the model parameter identification value b1 ′ is equal to or less than the upper limit value X_IDB1H calculated in step 83 of FIG. . When the determination result is YES and X_IDB1L ≦ b1 ′ ≦ X_IDB1H, the process proceeds to step 132, the model parameter b1 is set to the identification value b1 ′, and the process is terminated.

  On the other hand, if the decision result in the step 131 is NO and b1 ′> X_IDB1H, the process proceeds to a step 133, and the model parameter b1 is set to the upper limit value X_IDB1H, and at the same time, the flag F_B1LMT is set to “1” to represent it. This process is terminated.

  On the other hand, if the decision result in the step 130 is NO and b1 ′ <X_IDB1L, the process proceeds to a step 134 where the model parameter b1 is set to the lower limit value X_IDB1L and at the same time the flag F_B1LMT is set to “1” to represent it. This process is terminated.

  By executing the above b1 ′ limit process, the model parameter b1 can be limited to a value within the restricted range of X_IDB1L and X_IDB1H, thereby preventing the occurrence of the drift phenomenon by the sequential identification algorithm. Can be avoided. Further, as described above, since these upper and lower limit values X_IDB1H and X_IDB1L are set according to the exhaust gas volume AB_SV, the restriction range is accompanied by a change in the operating state of the engine 3, that is, a change in the dynamic characteristics of the controlled object. It can be set as an appropriate range of the stability limit reflecting the change of the stability limit, and the stability of the control system can be ensured by using the model parameter b1 regulated within such a regulation range. it can. As described above, the stability of the control system can be improved, and the post-catalyst exhaust gas characteristics can be improved.

  Next, the calculation processing of the state predictor 22 in step 33 described above will be described with reference to FIG. In this process, first, in step 140, the matrix elements α1, α2, βi, βj of the above-described equation (7) are calculated. Next, the process proceeds to step 141, and the matrix element α1, α2, βi, βj calculated in step 140 is applied to the equation (7) to calculate the predicted value PREVO2 of the output deviation VO2, and then the present process is terminated.

  Next, the process for calculating the control amount Usl in step 34 described above will be described with reference to FIG. In this process, first, in step 150, the prediction switching function σPRE is calculated by the aforementioned equation (38) in FIG.

Next, proceeding to step 151, the integrated value SUMSIGMA of the prediction switching function σPRE is calculated. In this process, as shown in FIG. 31, first, in step 160, it is determined whether or not at least one of the following three conditions (l) to (n) is satisfied.
(L) The adaptive control flag F_PRISMON is “1”.
(M) An accumulated value holding flag F_SS_HOLD described later is “0”.
(N) An ADSM execution flag F_KOPR described later is “0”.

  When the determination result in step 160 is YES, that is, when the calculation condition of the integrated value SUMSIGMA is satisfied, the process proceeds to step 161, where the current value SUMSIGMA (k) of the integrated value SUMSIGMA is changed to the previous value SUMSIGMA (k-1). Is set to a value obtained by adding the product of the control cycle ΔT and the prediction switching function σPRE [SUMSIGMA (k−1) + ΔT · σPRE].

  Next, the routine proceeds to step 162, where it is determined whether or not the current value SUMSIGMA (k) calculated at step 161 is greater than a predetermined lower limit value SUMSL. When the determination result is YES, the process proceeds to step 162 to determine whether or not the current value SUMSIGMA (k) is smaller than a predetermined upper limit value SUMSH. When the determination result is YES and SUMSL <SUMMSIGMA (k) <SUMSH, this processing is ended as it is.

  On the other hand, if the determination result in step 163 is NO and SUMSIGMA (k) ≧ SUMSH, the process proceeds to step 164, and the current value SUMSIGMA (k) is set to the upper limit value SUMSH, and then the present process ends. On the other hand, if the determination result in step 162 is NO and SUMSIGMA (k) ≦ SUMSL, the process proceeds to step 165 to set the current value SUMSIGMA (k) to the lower limit value SUMSL, and then the present process is terminated.

  On the other hand, when the determination result of step 160 is NO, that is, when all of the three conditions (l) to (n) are not satisfied and the calculation condition of the integrated value SUMSIGMA is not satisfied, the process proceeds to step 166 and the current value SUMSIGMA ( k) is set to the previous value SUMSIGMA (k-1). That is, the integrated value SUMSIGMA is held. Then, this process is complete | finished.

  Returning to FIG. 30, in steps 152 to 154 following step 151, the equivalent control input Ueq, the reaching law input Urch, and the adaptive law input Uadp are calculated by the above-described equations (40) to (42) of FIG. 12.

  Next, the process proceeds to step 155, and the sum of these equivalent control input Ueq, reaching law input Urch and adaptive law input Uadp is set as the control amount Usl, and then this process is terminated.

  Next, the calculation process of the sliding mode control amount DKCMDSLD in step 36 of FIG. 15 will be described with reference to FIGS. In this process, first, in step 170, a limit value calculation process for the control amount Usl is executed. The details of this process are omitted, but based on the determination result of the controller stability determination process in step 35 described above and adaptive upper and lower limit values Usl_ah, Usl_al of the control amount Usl described later, for non-idle operation Upper and lower limit values Usl_ahf, Usl_alf, and upper and lower limit values Usl_ahfi, Usl_alfi for idle operation are calculated.

  Next, the routine proceeds to step 171 where it is determined whether or not the idle operation flag F_IDLE is “0”. If the determination result is YES and the engine is not idling, the process proceeds to step 172 to determine whether or not the control amount Usl calculated in the above-described processing of FIG. .

  If this determination result is NO and Usl> Usl_alf, the routine proceeds to step 173, where it is determined whether or not the control amount Usl is not less than the upper limit value Usl_ahf for non-idle operation. If the determination result is NO and Usl_alf <Usl <Usl_ahf, the process proceeds to step 174 where the sliding mode control amount DKCMDSLD is set to the control amount Usl and at the same time the integrated value holding flag F_SS_HOLD is set to “0”.

  Next, the process proceeds to step 175, where the current value Usl_al (k) of the adaptation lower limit value is set to a value [Usl_al (k-1) + X_AL_DEC] obtained by adding a predetermined decrease side value X_AL_DEC to the previous value Usl_al (k-1). At the same time, after setting the current value Usl_ah (k) of the adaptation upper limit value to a value [Usl_al (k-1) -X_AL_DEC] obtained by subtracting a predetermined decrease side value X_AL_DEC from the previous value Usl_ah (k-1). Exit.

  On the other hand, if the determination result in step 173 is YES and Usl ≧ Usl_ahf, the process proceeds to step 176, where the sliding mode control amount DKCMDSLD is set to the adaptive upper limit value Usl_ahf for non-idle operation, and at the same time, the integrated value holding flag F_SS_HOLD is set to “1”. Set to "".

  Next, the routine proceeds to step 177, where it is determined whether or not the timer value TMACR of the post-start timer is smaller than the predetermined time X_TMAWAST or that the post-F / C determination flag F_AFC is “1”. The after-start timer is an up-count timer that measures the elapsed time after the engine 3 is started.

  When the determination result is YES, that is, when the predetermined time X_TMAWAST has not elapsed since the engine was started or when the predetermined time X_TM_AFC has not elapsed after the end of the fuel cut operation, the present process is terminated.

  On the other hand, when the determination result in step 177 is NO, that is, when the predetermined time X_TMAWAST has elapsed after the engine is started and the predetermined time X_TM_AFC has elapsed after the fuel cut operation ends, the routine proceeds to step 178, where the adaptive lower limit value is set. The current value Usl_al (k) is set to a value [Usl_al (k-1) + X_AL_DEC] obtained by adding the decreasing value X_AL_DEC to the previous value Usl_al (k-1), and at the same time, the current value Usl_ah (k) of the adaptation upper limit value is set. Then, after setting to the value [Usl_ah (k−1) + X_AL_INC] obtained by adding a predetermined increase side value X_AL_INC to the previous value Usl_ah (k−1), the present process is terminated.

  On the other hand, if the determination result in step 172 is YES and Usl ≦ Usl_alf, the process proceeds to step 179, and the sliding mode control amount DKCMDSLD is set to the adaptive lower limit Usl_alf for non-idle operation, and at the same time, the accumulated value holding flag F_SS_HOLD is set to “1”. Set to "".

  Next, the routine proceeds to step 180, where it is determined whether or not the second start flag F_VST is “1”. If the determination result is YES, and the second predetermined time TVST has not elapsed since the vehicle started and the vehicle is in the second start mode, the present process is terminated.

  On the other hand, when the determination result in step 180 is NO and the second predetermined time TVST has elapsed after the start of the vehicle and the second start mode ends, the process proceeds to step 181 where the current value Usl_al (k) of the adaptation lower limit value is set. Are set to a value [Usl_al (k−1) −X_AL_INC] obtained by subtracting the increasing side value X_AL_INC from the previous value Usl_al (k−1), and at the same time, the current value Usl_ah (k) of the adaptation upper limit value is set to the previous value Usl_ah (k). -1) is set to a value obtained by subtracting the decreasing value X_AL_DEC from [Usl_ah (k-1) -X_AL_DEC]. Thereafter, this process is terminated.

  On the other hand, if the determination result in step 171 is NO and the engine is in idle operation, the process proceeds to step 182 in FIG. 33, and it is determined whether or not the control amount Usl is equal to or less than the lower limit value Usl_alfi for idle operation. When the determination result is NO and Usl> Usl_alfi, the process proceeds to step 183 to determine whether or not the control amount Usl is equal to or greater than the upper limit value Usl_ahfi for idle operation.

  If the determination result is NO and Usl_alfi <Usl <Usl_ahfi, the process proceeds to step 184, the sliding mode control amount DKCMDSLD is set to the control amount Usl, and at the same time, the integrated value holding flag F_SS_HOLD is set to “0”, and then this processing is performed. Exit.

  On the other hand, if the determination result in step 183 is YES and Usl ≧ Usl_ahfi, the process proceeds to step 185 to set the sliding mode control amount DKCMDSLD to the upper limit value Usl_ahfi for idle operation and simultaneously set the accumulated value holding flag F_SS_HOLD to “1”. After setting, this process is terminated.

  On the other hand, if the decision result in the step 182 is YES and Usl ≦ Usl_alfi, the process proceeds to a step 186 to set the sliding mode control amount DKCMDSLD to the lower limit value Usl_alfi for idle operation and at the same time set the accumulated value holding flag F_SS_HOLD to “1”. After setting, this process is terminated.

  Next, the process for calculating the ΔΣ modulation control amount DKCMDDSM in step 37 of FIG. 15 will be described with reference to FIG. As shown in the figure, in this process, first, in step 190, the current value DSMSGNS (k) [= u ″ (k)] of the DSM signal value calculated in the previous loop stored in the RAM. Is set as the previous value DSMSGNS (k−1) [= u ″ (k−1)].

Next, the process proceeds to step 191, where the current value DSMSIGMA (k) [= σ _d (k)] of the deviation integrated value calculated in the previous loop stored in the RAM is changed to the previous value DSMSIGMA (k−1). Set as [= σ _d (k−1)].

Next, the routine proceeds to step 192, where it is determined whether or not the output deviation predicted value PREVO2 (k) is greater than or equal to zero. When the determination result is YES, assuming that the engine 3 is in an operating state in which the air-fuel ratio of the air-fuel mixture should be changed to the lean side, the process proceeds to step 193 and the gain KRDSM (= G _d ) for the reference signal value is made lean. After the value KRDDSML is set, the process proceeds to step 195 described later.

  On the other hand, if the determination result in step 192 is NO, it is determined that the engine 3 is in an operating state in which the air-fuel ratio of the air-fuel mixture should be changed to the rich side, and the process proceeds to step 194 where the reference signal value gain KRDSM After setting the value for enrichment KRDSMR larger than the value KRDSML, the process proceeds to step 195.

  Thus, the reason why the leaning value KRDSML and the richening value KRDSMR are set to different values is as follows. That is, when the air-fuel ratio of the air-fuel mixture is changed to the lean side, the value for leaning is used to obtain the effect of suppressing the NOx emission amount due to the lean bias in order to ensure the NOx purification rate of the first catalytic device 8a. By setting KRDSML to a value smaller than the enrichment value KRDSMR, the air-fuel ratio is controlled so that the convergence speed of the output Vout of the O2 sensor 15 to the target value Vop is slower than when changing to the rich side. To do. On the other hand, when the air-fuel ratio of the air-fuel mixture is changed to the rich side, the rich value KRDSMR is set to the lean value in order to sufficiently recover the NOx purification rates of the first and second catalytic devices 8a and 8b. By setting the value larger than KRDSML, the air-fuel ratio is controlled so that the convergence speed of the output Vout of the O2 sensor 15 to the target value Vop becomes faster than when changing to the lean side. As described above, when the air-fuel ratio of the air-fuel mixture is changed to the rich side and the lean side, good post-catalyst exhaust gas characteristics can be ensured.

  In step 195 following step 193 or 194, the previous value of the DSM signal value calculated in step 190 is obtained by multiplying the value −1, the gain KRDSM for the reference signal value, and the current value PREVO2 (k) of the predicted value. A value obtained by subtracting DSMSGNS (k−1) [−1 · KRDSM · PREVO2 (k) −DSMSGNS (k−1)] is set as a deviation signal value DSMDELTA [= δ (k)]. This process corresponds to the aforementioned equations (27) and (28).

  Next, the process proceeds to step 196, where the current value DSMSIGMA (k) of the deviation integral value is the sum of the previous value DSMSIGMA (k−1) calculated in step 191 and the deviation signal value DSMDELTA calculated in step 195 [DSSMSIGMA (k -1) + DSMDELTA]. This process corresponds to the aforementioned equation (29).

  Next, in steps 197 to 199, when the current value DSMSIGMA (k) of the deviation integral value calculated in step 196 is greater than or equal to 0, the current value DSMSGNS (k) of the DSM signal value is set to the value 1, and the deviation integral is set. When the current value DSMSIGMA (k) of the value is smaller than the value 0, the current value DSMSGNS (k) of the DSM signal value is set to the value -1. The processes in steps 197 to 199 correspond to the above-described equation (30).

Next, at step 200, the gain KDSM (= F _d ) for the DSM signal value is calculated by searching the table shown in FIG. 35 according to the exhaust gas volume AB_SV. As shown in the figure, the gain KDSM is set to a larger value as the exhaust gas volume AB_SV is smaller. This is to compensate for the lower responsiveness of the output Vout of the O2 sensor 15 as the exhaust gas volume AB_SV is smaller, that is, as the operating load of the engine 3 is smaller. By setting the gain KSDM in this way, the ΔΣ modulation control amount DKCMDDSM can be appropriately calculated according to the operating state of the engine 3 while avoiding, for example, an over-gain state and the like. Gas characteristics can be improved.

  The table used for calculating the gain KDSM is not limited to the above table in which the gain KDSM is set in accordance with the exhaust gas volume AB_SV, but a parameter (for example, the basic fuel injection time Tim) that represents the operating load state of the engine 3. Accordingly, any gain KDSM may be set in advance. In the case where the deterioration discriminators of the catalyst devices 8a and 8b are provided, the gain DSM is corrected to a smaller value as the deterioration degree of the catalyst devices 8a and 8b determined by the deterioration discriminator is larger. It may be.

  Next, proceeding to step 201, the ΔΣ modulation control amount DKCMDDSM is set to a value [KDSM · DSMSGNS (k)] obtained by multiplying the gain KDSM for the DSM signal value by the current value DSMSGNS (k) of the DSM signal value. Then, this process is terminated. This process corresponds to the aforementioned equation (31).

  Next, the process for calculating the adaptive target air-fuel ratio KCMDLSD in step 38 of FIG. 15 will be described with reference to FIG. As shown in the figure, in this process, first, in step 210, it is established whether the idle operation flag F_IDLE is “1” and the idle ADSM execution flag F_SWOPRI is “1”. Determine whether or not. The idling ADSM execution flag F_SWOPPRI is set to “1” when the engine 3 is in an idling operation and is in an operating state in which ADSM processing is to be executed, and is set to “0” otherwise.

  When the determination result is YES, that is, when the engine 3 is in idling operation and is in an operation state in which the adaptive target air-fuel ratio KCMDLSD is to be calculated by ADSM processing, the process proceeds to step 211 and the adaptive target air-fuel ratio KCMDLSLD is set to the reference value FLAFBASE by ΔΣ A value [FLAFBASE + DKCMDDSM] obtained by adding the modulation control amount DKCMDDSM is set. This process corresponds to the aforementioned equation (32).

  Next, the process proceeds to step 212, in which the ADSM executed flag F_KOPR is set to “1” to indicate that the ADSM process has been executed, and then this process ends.

On the other hand, when the determination result of step 210 is NO, the process proceeds to step 213 to determine whether or not the catalyst / O2 sensor flag F_FCATDSM is “1”. The catalyst / O2 sensor flag F_FCATDSM is set to “1” when at least one of the following four conditions (o) to (r) is satisfied, and is set to “0” otherwise.
(O) The catalyst capacity of the first catalyst device 8a is not less than a predetermined value.
(P) The noble metal content of the first catalyst device 8a is not less than a predetermined value.
(Q) The LAF sensor 14 is not provided in the exhaust pipe 7 of the engine 3.
(R) The O2 sensor 15 is provided downstream of the second catalyst device 8b.

  When the determination result is YES, the process proceeds to step 214 to determine whether or not both the first start flag F_VOTVST and the post-start ADSM execution flag F_SWOPRVST are “1”. The post-start ADSM execution flag F_SWOPRVST is set to “1” after the start of the vehicle and when the engine 3 is in an operating state in which ADSM processing is to be executed, and is set to “0” otherwise.

  When the determination result is YES, that is, when the first predetermined time TVOTVST has elapsed after the vehicle has started and the ADSM process is to be executed, as described above, after executing steps 211 and 212, The process ends.

  On the other hand, if the determination result in step 214 is NO, the process proceeds to step 215, where both the exhaust gas volume AB_SV is equal to or smaller than the predetermined value OPRSVH and the small exhaust ADSM execution flag F_SWOPRSV is “1” are established. It is determined whether or not. The small exhaust time ADSM execution flag F_SWOPRSV is set to “1” when the exhaust gas volume AB_SV of the engine 3 is small and the engine 3 is in an operating state in which ADSM processing should be executed, and to “0” otherwise. Set.

  When the determination result is YES, that is, when the exhaust gas volume AB_SV is small and the engine 3 is in an operating state in which ADSM processing is to be executed, as described above, after executing steps 211 and 212, this processing is terminated. .

  On the other hand, when the determination result in step 215 is NO, it is determined that the engine 3 is in an operating state in which PRISM processing should be executed, and the process proceeds to step 216 where the adaptive target air-fuel ratio KCMDLSLD is set to the reference value FLAFBASE and the adaptive correction term FLAFADP and sliding mode. A value obtained by adding the control amount DKCMDSLD is set to [FLAFBASE + FLAFADP + DKCMDSLD]. Next, the process proceeds to step 217, and the ADSM executed flag F_KOPR is set to “0” to indicate that the PRISM process has been executed, and then this process ends.

  On the other hand, when the determination result in step 213 is NO, that is, when none of the above-described four conditions (o) to (r) is satisfied, steps 214 and 215 are skipped, and steps 216 and 217 described above are performed. After execution, this process is terminated. As described above, in the calculation process of the adaptive target air-fuel ratio KCMDLSD, the adaptive target air-fuel ratio KCMDLSLD is calculated by switching to the ADSM process or PRISM process according to the operating state of the engine 3.

  Next, the calculation process of the adaptive correction term FLAFADP in step 39 of FIG. 15 will be described with reference to FIG. As shown in the figure, in this process, first, in step 220, it is determined whether or not the output deviation VO2 is a value within a predetermined range (ADL <VO2 <ADH). When the determination result is YES, that is, when the output deviation VO2 is small and the output Vout of the O2 sensor 15 is in the vicinity of the target value Vop, the process proceeds to step 221, and whether or not the adaptive law input Uadp is smaller than the predetermined lower limit value NRL. Is determined.

  If this determination result is NO and Uadp ≧ NRL, the routine proceeds to step 222, where it is determined whether or not the adaptive law input Uadp is larger than a predetermined upper limit value NRH. If the determination result is NO and NRL ≦ Uadp ≦ NRH, the routine proceeds to step 223, where the current value FLAFADP (k) of the adaptive correction term is set to the previous value FLAFADP (k−1). That is, the value of the adaptive correction term FLAFADP is held. Then, this process is complete | finished.

  On the other hand, if the decision result in the step 222 is YES and Uadp> NRH, the process proceeds to a step 224 to add the current value FLAFADP (k) of the adaptive correction term and the predetermined update value X_FLAFDLT to the previous value FLAFADP (k−1). After this value [FLAFADP (k−1) + X_FLAFDLT] is set, this process is terminated.

  On the other hand, if the decision result in the step 221 is YES and Uadp <NRL, the process advances to a step 225 to subtract the current value FLAFADP (k) of the adaptive correction term and the predetermined update value X_FLAFDLT from the previous value FLAFADP (k−1). After this value [FLAFADP (k−1) −X_FLAFDLT] is set, the present process is terminated.

  As described above, according to the control device 1 of the first embodiment, the target air-fuel ratio KCMD is used as the control input, and the output Vout of the O2 sensor 15 is used as the output. In the control target, the control timing shift between the input and output of the control target can be properly eliminated, thereby improving the stability and controllability of the control and improving the exhaust gas characteristics after the catalyst Can be made.

  Hereinafter, control devices according to second to eighth embodiments of the present invention will be described. In the following description of each embodiment, the same or equivalent components as those in the first embodiment described above are denoted by the same reference numerals, and the description thereof will be omitted as appropriate.

  First, the control device of the second embodiment will be described with reference to FIG. As shown in the figure, the control device 1 of the second embodiment is different from the control device 1 of the first embodiment only in the on-board identifier 23. Specifically, the on-board identifier 23 of the first embodiment calculates the model parameters a1, a2, and b1 based on KACT, Vout, and φop (KCMD), while the on-board identifier of the present embodiment is on. In the board identifier 23, model parameters a1, a2, and b1 are calculated based on Vout and φop.

  That is, in the on-board identifier 23, the identification shown in the above-described equations (8) to (15) in FIG. 5 instead of the identification algorithm shown in the equations (16) to (23) in FIG. 6 of the first embodiment. The model parameter identification values a1 ′, a2 ′, and b1 ′ are calculated by the algorithm, and the model parameters a1, a2, and b1 are calculated by performing the limit processing shown in FIGS. Although a specific program for the arithmetic processing of the on-board identifier 23 is not shown, it is configured in substantially the same way as that of the first embodiment. According to the control device 1 of the present embodiment as described above, the same effects as those of the control device 1 of the first embodiment can be obtained.

  Next, a control device according to a third embodiment will be described with reference to FIG. As shown in the figure, the control device 1 of the third embodiment differs from the control device 1 of the first embodiment only in the state predictor 22. Specifically, in the state predictor 22 of the first embodiment, the predicted value PREVO2 is calculated based on a1, a2, b1, KACT, Vout, and φop (KCMD), whereas in the present embodiment, The on-board identifier 23 calculates a predicted value PREVO2 based on a1, a2, b1, Vout and φop.

  That is, in this state predictor 22, the predicted value PREVO2 of the output deviation VO2 is obtained by the prediction algorithm shown in the equation (6) in FIG. 4 instead of the prediction algorithm shown in the equation (7) in FIG. 4 of the first embodiment. Calculated. Although a specific program for the arithmetic processing of the state predictor 22 is not illustrated, it is configured in substantially the same manner as that of the first embodiment. According to this control device 1, the same effect as that of the control device 1 of the first embodiment can be obtained.

  Next, a control device according to a fourth embodiment will be described with reference to FIG. As shown in the figure, the control device 1 of the fourth embodiment is different from the control device 1 of the first embodiment in that it replaces the ADSM controller 20, the PRISM controller 21 and the onboard identifier 23 with a schedule type DSM. The only difference is that the model parameters a1, a2, and b1 are calculated by using the controller 20A, the schedule type state prediction sliding mode controller 21A, and the parameter scheduler 28 (model parameter setting means).

  In the parameter scheduler 28, first, the exhaust gas volume AB_SV is calculated based on the engine speed NE and the intake pipe absolute pressure PBA by the above-described equation (44). Next, model parameters a1, a2, and b1 are calculated according to the exhaust gas volume AB_SV by the table shown in FIG.

  In this table, the model parameter a1 is set to a smaller value as the exhaust gas volume AB_SV is larger. Conversely, the model parameters a2 and b1 are larger values as the exhaust gas volume AB_SV is larger. Is set to This is because the output to be controlled, that is, the output Vout of the O2 sensor 15 is stabilized as the exhaust gas volume AB_SV increases, while the output Vout of the O2 sensor 15 becomes oscillating as the exhaust gas volume AB_SV decreases. by.

  The schedule type DSM controller 20A uses the model parameters a1, a2, and b1 calculated as described above to calculate the target air-fuel ratio KCMD by the DSM controller 24 similar to that of the first embodiment described above. The schedule type state prediction sliding mode controller 21A also uses the model parameters a1, a2 and b1 calculated as described above to calculate the target air-fuel ratio KCMD by the SLD controller 25 similar to that of the first embodiment described above.

  According to this control device 1, the same effect as that of the control device 1 of the first embodiment can be obtained. In addition, by using the parameter scheduler 28, the model parameters a1, a2, and b1 can be calculated more quickly than when the on-board identifier 23 is used. Thereby, the control responsiveness can be improved, and good post-catalyst exhaust gas characteristics can be secured more quickly.

  Next, a control device according to a fifth embodiment will be described with reference to FIG. The control device 1 of the fifth embodiment is different only in that an SDM controller 29 is used instead of the DSM controller 24 of the control device 1 of the first embodiment. The SDM controller 29 calculates the control input φop (k) based on the predicted value PREVO2 (k) by a control algorithm to which the ΣΔ modulation algorithm is applied.

That is, as shown in the figure, in the SDM controller 29, an inverting amplifier 29a, the reference signal r (k), the value -1, the gain G _d for reference signal and the predicted value PREVO2 (k) of the multiplication to each other Generated as a signal. Next, the integrator 29b converts the reference signal integrated value σ _d r (k) to the sum of the reference signal integrated value σ _d r (k−1) delayed by the delay element 29c and the reference signal r (k). Generated as a signal. On the other hand, an integrator 29d, SDM signal integrated value σ _d u (k) is a delay element and SDM signal integrated value σ _d u (k-1) delayed by 29e, SDM signal u which is delayed by a delay element 29j '' (k-1) and a sum signal. Then, a difference signal δ ″ (k) between the reference signal integrated value σ _d r (k) and the SDM signal integrated value σ _d u (k) is generated by the differencer 29f.

Next, the SDM signal u ″ (k) is generated as a value obtained by encoding the deviation signal δ ″ (k) by the quantizer 29g (sign function). Then, the amplified SDM signal u (k) is generated as a value obtained by amplifying the SDM signal u ″ (k) with a predetermined gain F _d by the amplifier 29h, and then this amplified SDM signal u ( The control input φop (k) is generated as a value obtained by adding k) to a predetermined reference value FLAFBASE.

The control algorithm of the above SDM controller 29 is expressed by the following equations (45) to (51).
r (k) = − 1 · G _d · PREVO2 (k) (45)
σ _d r (k) = σ _d r (k−1) + r (k) (46)
σ _d u (k) = σ _d u (k−1) + u ″ (k−1) (47)
δ ″ (k) = σ _d r (k) −σ _d u (k) (48)
u ″ (k) = sgn (δ ″ (k)) (49)
u (k) = F _d · u ″ (k) (50)
φop (k) = FLAFBASE + u (k) (51)
Here, G _d and F _d represent gains. The value of the sign function sgn (δ ″ (k)) is sgn (δ ″ (k)) = 1 when δ ″ (k) ≧ 0, and when δ ″ (k) <0. sgn (δ ″ (k)) = − 1 (Note that when δ ″ (k) = 0, sgn (δ ″ (k)) = 0 may be set).

  The characteristics of the ΣΔ modulation algorithm in the control algorithm of the SDM controller 29 described above are similar to those of the ΔΣ modulation algorithm. When the SDM signal u (k) is input to the control target, the reference signal r (k) is the control target. This is because it can be generated (calculated) as a value that is reproduced in the output. That is, the SDM controller 29 has a characteristic that it can generate the control input φop (k) similar to the DSM controller 24 described above. Therefore, according to the control device 1 of this embodiment using this SDM controller 29, the same effect as that of the control device 1 of the first embodiment can be obtained. A specific program for the SDM controller 29 is not shown, but is configured in substantially the same manner as the DSM controller 24.

  Next, a control device according to a sixth embodiment will be described with reference to FIG. The control device 1 of the sixth embodiment is different only in that a DM controller 30 is used instead of the DSM controller 24 of the control device 1 of the first embodiment. The DM controller 30 calculates a control input φop (k) based on the predicted value PREVO2 (k) by a control algorithm to which a Δ modulation algorithm is applied.

That is, as shown in the figure, in this DM controller 30, the reference signal r (k) is multiplied by the value −1, the reference signal gain G _d, and the predicted value PREVO2 (k) by the inverting amplifier 30a. Generated as a signal. On the other hand, an integrator 30b, DM signal integrated value σ _d u (k) is a DM signal integrated value delayed by a delay element _{30c σ d u (k-1} ), the delay elements DM signal delayed by 30h u '' (k-1) and a sum signal. Then, a difference signal δ ″ (k) between the reference signal r (k) and the DM signal integrated value σ _d u (k) is generated by the differencer 30d.

Next, the DM signal u ″ (k) is generated as a value obtained by encoding the deviation signal δ ″ (k) by the quantizer 30e (sign function). The amplified DM signal u (k) is generated by the amplifier 30f as a value obtained by amplifying the DM signal u ″ (k) with a predetermined gain F _d , and then this amplified DM signal u ( The control input φop (k) is generated as a value obtained by adding k) to a predetermined reference value FLAFBASE.

The control algorithm of the DM controller 30 described above is expressed by the following equations (52) to (57).
r (k) = − 1 · G _d · PREVO2 (k) (52)
σ _d u (k) = σ _d u (k−1) + u ″ (k−1) (53)
δ ″ (k) = r (k) −σ _d u (k) (54)
u ″ (k) = sgn (δ ″ (k)) (55)
u (k) = F _d · u ″ (k) (56)
φop (k) = FLAFBASE + u (k) (57)
Here, G _d and F _d represent gains. The value of the sign function sgn (δ ″ (k)) is sgn (δ ″ (k)) = 1 when δ ″ (k) ≧ 0, and when δ ″ (k) <0. sgn (δ ″ (k)) = − 1 (Note that when δ ″ (k) = 0, sgn (δ ″ (k)) = 0 may be set).

  The characteristics of the control algorithm, that is, the Δ modulation algorithm of the DM controller 30 described above are similar to the ΔΣ modulation algorithm and the ΣΔ modulation algorithm when the DM signal u (k) is input to the control target, and the reference signal r (k) is the control target. The DM signal u (k) can be generated (calculated) as a value that is reproduced in the output. That is, the DM controller 30 has a characteristic that it can generate the control input φop (k) similar to the DSM controller 24 and the SDM controller 29 described above. Therefore, according to the control apparatus 1 of this embodiment using this DM controller 30, the same effect as the control apparatus 1 of 1st Embodiment can be acquired. A specific program of the DM controller 30 is not shown, but is configured in substantially the same manner as the DSM controller 24.

  Next, a control device according to a seventh embodiment will be described with reference to FIGS. 44 and 45. As shown in FIG. 44, the control device 1 of the seventh embodiment is not provided with the LAF sensor 14 in the engine 3 and the O2 sensor 15 is the second catalyst as compared with the control device 1 of the first embodiment. Only the point provided on the downstream side of the device 8b is different.

  In addition, since the LAF sensor 14 is not provided, in the control device 1, as shown in FIG. 45, the on-board identifier 23 uses the output Vout of the O2 sensor 15 and the control input φop (target air-fuel ratio KCMD). Model parameters a1, a2 and b1 are calculated. That is, the on-board identifier 23 calculates model parameter identification values a1 ′, a2 ′, b1 ′ by the above-described identification algorithm shown in the equations (8) to (15) in FIG. By performing the limit processing described above, model parameters a1, a2, and b1 are calculated.

  Further, the state predictor 22 calculates a predicted value PREVO2 of the output deviation VO2 based on the model parameters a1, a2, b1, the output Vout of the O2 sensor 15 and the control input φop. That is, in this state predictor 22, the predicted value PREVO2 of the output deviation VO2 is calculated by the prediction algorithm shown in the equation (6) in FIG. Although specific programs for the arithmetic processing of the state predictor 22 and the on-board identifier 23 are not shown, they are configured in substantially the same manner as in the first embodiment, and other programs are also implemented in the first embodiment. The configuration is the same as that of the form.

  According to the control device 1 of the present embodiment as described above, the same effects as those of the control device 1 of the first embodiment can be obtained. In particular, as described above, in steps 192 to 194 in FIG. 34, the reference signal value gain KRDSM is set to a different value between when the exhaust gas is controlled to the lean side and when it is controlled to the rich side. Even when the air-fuel ratio is controlled only by the O2 sensor 15 as in the present embodiment by changing the convergence speed of the target air-fuel ratio KCMD to the target value Vop, the air-fuel ratio of the air-fuel mixture is set to the rich side and the lean side. When changing to, good post-catalyst exhaust gas characteristics can be reliably obtained. In addition to this, good post-catalyst exhaust gas characteristics can be ensured without using the LAF sensor 14, and thus the manufacturing cost can be reduced accordingly.

  Next, a control device according to an eighth embodiment will be described with reference to FIG. As shown in the figure, the control device 1 of the eighth embodiment is similar to the control device 1 of the seventh embodiment in that the ADSM controller 20, PRISM controller 21 and onboard identifier 23 are the same as those in the fourth embodiment. The schedule type DSM controller 20A, the schedule type state prediction sliding mode controller 21A and the parameter scheduler 28 are replaced, and these controllers 20A and 21A and the parameter scheduler 28 are configured in the same manner as in the fourth embodiment. . According to the control device 1, the same effect as that of the control device 1 of the seventh embodiment can be obtained. In addition, by using the parameter scheduler 28, the model parameters a1, a2, and b1 can be calculated more quickly than when the on-board identifier 23 is used. Thereby, the control responsiveness can be improved, and good post-catalyst exhaust gas characteristics can be secured more quickly.

  In addition, although each above embodiment is an example which comprised the control apparatus of this invention as what controls the air fuel ratio of the internal combustion engine 3, this invention is not limited to this, Other arbitrary control objects are controlled. Needless to say, the present invention is widely applicable to control devices. Further, the ADSM controller 20 and the PRISM controller 21 may be configured by an electric circuit instead of the program of the embodiment.

1 is a diagram illustrating a schematic configuration of a control device according to a first embodiment of the present invention and an internal combustion engine to which the control device is applied. FIG. Results of measuring the purification rate of HC and NOx of both first catalyst devices and the output Vout of the O2 sensor 15 with respect to the output KACT of the LAF sensor when the first catalyst device in the deteriorated state and the undegraded state is used It is a figure which shows an example. It is a block diagram which shows the structure of the ADSM controller and PRISM controller of the control apparatus of 1st Embodiment. It is a figure which shows an example of the numerical formula of the prediction algorithm of a state predictor. It is a figure which shows an example of the numerical formula of the identification algorithm of an on-board identifier. It is a figure which shows another example of the numerical formula of the identification algorithm of an on-board identifier. It is a block diagram which shows the structure of the controller which performs delta-sigma modulation, and a control system provided with the same. It is a timing chart which shows an example of the control result of the control system of FIG. 3 is a timing chart for explaining the principle of adaptive prediction type ΔΣ modulation control by the ADSM controller of the first embodiment. It is a block diagram which shows the structure of the DSM controller among ADSM controllers. It is a figure which shows the numerical formula of a sliding mode control algorithm. It is a figure which shows the numerical formula of the sliding mode control algorithm of a PRISM controller. It is a flowchart which shows the fuel-injection control process of an internal combustion engine. It is a flowchart which shows an adaptive air fuel ratio control process. It is a flowchart which shows the continuation of FIG. It is a flowchart which shows the start determination process in step 21 of FIG. It is a flowchart which shows the execution determination process of PRISM / ADSM process in step 23 of FIG. It is a flowchart which shows the execution determination process of an identifier operation in step 24 of FIG. It is a flowchart which shows the calculation process of the various parameters in step 25 of FIG. It is a figure which shows an example of the table used for calculation of dead time CAT_DELAY, KACT_D. It is a figure which shows an example of the table used for calculation of weight parameter (lambda) 1. It is a figure which shows an example of the table used for calculation of limit value X_IDA2L, X_IDB1L, and X_IDB1H which restrict | limit the value of model parameter a1, a2, b1. It is a figure which shows an example of the table used for calculation of filter order n. It is a flowchart which shows the calculation process of the identifier in step 31 of FIG. It is a flowchart which shows the stabilization process of (k) in step 94 of FIG. It is a flowchart which shows the limit process of a1 '& a2' in step 101 of FIG. It is a figure which shows the control range in which the combination of a1 '& a2' is controlled by the process of FIG. It is a flowchart which shows the limit process of b1 'in step 102 of FIG. It is a flowchart which shows the calculation process of the state predictor of step 33 of FIG. It is a flowchart which shows the calculation process of the control amount Usl of step 34 of FIG. FIG. 31 is a flowchart showing an integrated value calculation process of a prediction switching function σPRE in step 151 of FIG. 16 is a flowchart showing a calculation process of a sliding mode control amount DKCMDSLD in step 36 of FIG. It is a flowchart which shows the continuation of FIG. 16 is a flowchart showing a calculation process of a ΔΣ modulation control amount DKCMDDSM in step 37 of FIG. 15. It is a figure which shows an example of the table used for calculation of KDSM. 16 is a flowchart showing a calculation process of an adaptive target air-fuel ratio KCMDDSLD in step 38 of FIG. It is a flowchart which shows the calculation process of the adaptive correction term FLAFADP of step 39 of FIG. It is a block diagram which shows schematic structure of the control apparatus of 2nd Embodiment. It is a block diagram which shows schematic structure of the control apparatus of 3rd Embodiment. It is a block diagram which shows schematic structure of the control apparatus of 4th Embodiment. It is a figure which shows an example of the table used for calculation of a model parameter in the parameter scheduler of the control apparatus of 4th Embodiment. It is a block diagram which shows schematic structure of the SDM controller of the control apparatus of 5th Embodiment. It is a block diagram which shows schematic structure of DM controller of the control apparatus of 6th Embodiment. It is a figure which shows schematic structure of the control apparatus which concerns on 7th Embodiment, and the internal combustion engine to which this is applied. It is a block diagram which shows the structure of the control apparatus of 7th Embodiment. It is a block diagram which shows the structure of the control apparatus of 8th Embodiment.

Explanation of symbols

1 Control device
2 ECU (predicted value calculation means, control input calculation means, gain parameter detection
Exit means, gain setting means, air-fuel ratio calculating means, operating state detecting means, medium
During value calculating means, target air-fuel ratio calculating means, multiplication means, the correction factor setting hand stage
)
3 Internal combustion engine
7 Exhaust pipe (exhaust passage)
8a First catalyst device (catalyst)
8b Second catalyst device (catalyst)
11 Intake pipe absolute pressure sensor (gain parameter detection means, operating state detection
Hand stage)
13 crank angle sensor (gain parameter detecting means, operating condition-detecting hand stage
)
14 LAF sensor (upstream air-fuel ratio sensor)
15 Oxygen concentration sensor (air-fuel ratio sensor, downstream air-fuel ratio sensor)
20 ADSM controller (control input calculation means)
22 State predictor (predicted value calculation means)
Gain for G _d reference signal (correction coefficient)
Gain for KRDSM reference signal value (Correction factor)
NE Engine speed (parameter that represents the operating state)
PBA Absolute pressure in the intake pipe (parameter indicating operating condition)
AB_SV Exhaust gas volume (parameters representing the dynamic characteristics of the controlled object, gain parameters)
Output of KACT LAF sensor (reflects the control input input to the control target
value)
DKACT LAF output deviation (represents the deviation between the value reflecting the control input input to the control target and the reference value, the value reflecting the control input input to the control target)
value)
KCMD target air-fuel ratio (control input)
DKCMD air-fuel ratio difference (deviation between the value representing the control input, the control input and the reference value)
FLAFBASE standard value KCMDLSLD Adaptive target air-fuel ratio (control input, target air-fuel ratio)
DSMSGNS DSM signal value (intermediate value, intermediate value of target air-fuel ratio)
KDSM gain DKCMDDSM ΔΣ modulation control amount (second intermediate value)
Vout Oxygen concentration sensor output (control target output)
Vop target value
VO2 output deviation PREVO2 predicted value of the output deviation
During the time dt prediction

Claims (14)

Based on a prediction algorithm, predicted value calculation means for calculating a predicted value of an output deviation that is a deviation between the output of the controlled object and a predetermined target value ;
Multiplying means for multiplying the calculated predicted value of the output deviation by a correction coefficient;
Correction coefficient setting means for setting the correction coefficient to a smaller value than when the predicted value of the output deviation is greater than or equal to a predetermined value; and
Based on a predicted value of the output deviation multiplied by the correction coefficient based on any one of a Δ modulation algorithm, a ΔΣ modulation algorithm, and a ΣΔ modulation algorithm, the output of the control target is set to a predetermined target value. Control input calculating means for calculating a control input to the control object for controlling to converge to
A control device comprising: The predicted value calculating means, based on said prediction algorithm, at least one of the value reflecting the the control input inputted to said controlled object and the calculated control input, and in accordance with the output of said controlled object, said output The control device according to claim 1, wherein a predicted value of the deviation is calculated. The prediction algorithm is an algorithm based on a controlled object model having a value representing one of the control input and a value reflecting the control input input to the controlled object, and the output deviation as a variable. The control device according to claim 1 or 2. A value representing one of the control input and a value reflecting the control input input to the control target reflects a deviation between the control input and a predetermined reference value, and the control input input to the control target. The control device according to claim 3, wherein the control device is one of deviations between a value and the predetermined reference value. The control input calculating means calculates an intermediate value according to the predicted value of the output deviation multiplied by the correction coefficient based on the one modulation algorithm, and multiplies the calculated intermediate value by a predetermined gain. 5. The control device according to claim 1, wherein the control input is calculated based on the obtained value. Gain parameter detection means for detecting a gain parameter representing the gain characteristic of the control object;
A gain setting means for setting a value of the predetermined gain according to the detected gain parameter;
The control device according to claim 5, further comprising: The control input calculation means calculates a second intermediate value according to the predicted value of the output deviation multiplied by the correction coefficient based on the one modulation algorithm, and the calculated second intermediate value The control device according to claim 1, wherein the control input is calculated by adding a predetermined value to. The predicted value calculation means calculates a predicted time from when the control input is input to the controlled object until it is reflected in the output of the controlled object according to the dynamic characteristics of the controlled object, and the calculated value. 8. The control device according to claim 1, wherein a predicted value of the output deviation is calculated according to the predicted time. The output of the control object is an output of a downstream air-fuel ratio sensor that is disposed downstream of the catalyst in the exhaust passage of the internal combustion engine and detects the air-fuel ratio of the exhaust gas after passing through the catalyst,
The control input to the control object is a target air-fuel ratio of the air-fuel mixture supplied to the internal combustion engine,
A value that reflects the control input that is input to the control target is disposed upstream of the catalyst in the exhaust passage and is detected by an upstream air-fuel ratio sensor that detects the air-fuel ratio of the exhaust gas before passing through the catalyst. Output,
Based on the prediction algorithm, the predicted value calculation means determines at least one of the target air-fuel ratio of the air-fuel mixture supplied to the internal combustion engine and the output of the upstream air-fuel ratio sensor, and the output of the downstream air-fuel ratio sensor. Accordingly, a predicted value of the output deviation is calculated,
The control input calculating means is configured to converge the output of the downstream air-fuel ratio sensor to the predetermined target value according to the predicted value of the output deviation multiplied by the correction coefficient based on the one modulation algorithm. The control apparatus according to claim 2, comprising an air-fuel ratio calculating means for calculating a target air-fuel ratio of an air-fuel mixture to be supplied to the internal combustion engine. An operation state detecting means for detecting an operation state of the internal combustion engine;
According to the detected operating state of the internal combustion engine, the predicted value calculation means is a period from when the air-fuel mixture of the target air-fuel ratio is supplied to the internal combustion engine until it is reflected in the output of the downstream air-fuel ratio sensor. The control device according to claim 9, wherein a prediction time is calculated, and a predicted value of the output deviation is further calculated in accordance with the calculated prediction time. An operation state detecting means for detecting an operation state of the internal combustion engine;
The air-fuel ratio calculating means includes
An intermediate value calculating means for calculating an intermediate value of a target air-fuel ratio of an air-fuel mixture to be supplied to the internal combustion engine according to a predicted value of the output deviation multiplied by the correction coefficient based on the one modulation algorithm;
Gain setting means for setting a gain according to the detected operating state of the internal combustion engine;
Target air-fuel ratio calculating means for calculating a target air-fuel ratio of an air-fuel mixture to be supplied to the internal combustion engine based on a value obtained by multiplying the calculated intermediate value by the set gain;
The control device according to claim 9, further comprising: The output of the control target is an output of an air-fuel ratio sensor that is disposed downstream of the catalyst in the exhaust passage of the internal combustion engine and detects the air-fuel ratio of the exhaust gas after passing through the catalyst.
The control input to the control object is a target air-fuel ratio of the air-fuel mixture supplied to the internal combustion engine,
The predicted value calculation means calculates a predicted value of the output deviation according to a target air-fuel ratio of an air-fuel mixture supplied to the internal combustion engine and an output of the air-fuel ratio sensor based on the prediction algorithm,
The control input calculating unit is configured to converge the output of the air-fuel ratio sensor to the predetermined target value based on the predicted value of the output deviation multiplied by the correction coefficient based on the one modulation algorithm. 3. The control apparatus according to claim 2, further comprising air-fuel ratio calculating means for calculating a target air-fuel ratio of an air-fuel mixture to be supplied to the internal combustion engine. An operation state detecting means for detecting an operation state of the internal combustion engine;
According to the detected operating state of the internal combustion engine, the predicted value calculation means is a predicted time from when the air-fuel mixture of the target air-fuel ratio is supplied to the internal combustion engine until it is reflected in the output of the air-fuel ratio sensor The control device according to claim 12, wherein a predicted value of the output deviation is calculated further according to the calculated predicted time. An operation state detecting means for detecting an operation state of the internal combustion engine;
The air-fuel ratio calculating means includes
An intermediate value calculating means for calculating an intermediate value of a target air-fuel ratio of an air-fuel mixture to be supplied to the internal combustion engine according to a predicted value of the output deviation multiplied by the correction coefficient based on the one modulation algorithm;
Gain setting means for setting a gain according to the detected operating state of the internal combustion engine;
Target air-fuel ratio calculating means for calculating a target air-fuel ratio of an air-fuel mixture to be supplied to the internal combustion engine based on a value obtained by multiplying the calculated intermediate value by the set gain;
The control device according to claim 12, comprising:

Images