JPH07166936A - Air-fuel ratio controller of internal combustion engine - Google Patents

Abstract

PURPOSE:To improve control reliability by forcibly executing learning at the time of appearance of a specified condition, regardless of output of an oxygen sensor on the downstream side, when output of the oxygen sensor on the downstream side of a three-way catalyst allows to learn a control coefficient at reversing timing. CONSTITUTION:Oxygen sensors 19 are provided respectively on upstream and downstream sides of a three-way catalyst equipped on the exhaust pipe 7 of an internal combustion engine 1, and a control coefficient necessary for air-fuel ratio control is calculated by a control constant calculation means in response to the output of the downstream side sensor 119. The calculated control constant is learnt by a learning means by a timing in which the output of the downstream side sensor 119 is reversed between rich and lean conditions. The air-fuel ratio control is executed by the air-fuel ratio control means based on the output of the upstream sides sensor 19 and the learnt control constant. In this case, learning is forcibly executed by a forcible learning means at the time of appearance of a specified condition, even if reverse is not generated in the output of the downstream side sensor 119, and thereby such a problem that a control cycle is long and responsiveness is slow is solved.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an air-fuel ratio control system for an internal combustion engine, which has oxygen sensors provided above and below the three-way catalyst in an exhaust pipe.

[0002]

2. Description of the Related Art Conventionally, an oxygen sensor is provided on the upstream side of a catalyst, and an air-fuel ratio control for controlling the air-fuel ratio near the stoichiometric air-fuel ratio according to the output signal of this oxygen sensor (hereinafter referred to as "upstream sensor"). In the apparatus, there is a system in which an oxygen sensor (hereinafter, referred to as “downstream sensor”) is further provided downstream of the catalyst. In such a system, by correcting the control constants (integral constant, skip amount, delay time, comparison voltage, etc.) of the air-fuel ratio control according to the output signal of the downstream sensor, variations in upstream sensor characteristics, deterioration over time, engine deterioration, etc. This suppresses the deterioration / variation of the exhaust emission due to the variation of the above (commonly known as "2O2 sensor system", for example, the technology described in Japanese Patent Laid-Open No. 62-60941).

Since this 2O 2 sensor system can monitor the air-fuel ratio downstream of the catalyst, does the actual air-fuel ratio fall within the catalyst window (a region where NOx, CO, and HC in exhaust gas are all decreased)? Whether or not it can be detected. As a result, the air-fuel ratio can be reliably controlled within the catalyst window, and the exhaust gas purification performance can be improved. For example, in a system having only upstream sensors, the exhaust gas of a specific cylinder is mainly monitored depending on the mounting position of the sensor, which may cause a situation such as "total exhaust gas does not enter the catalyst window." According to the 2O2 sensor system, such a problem can be solved, and further, it becomes possible to optimize the exhaust gas purification performance against not only the variation of the upstream sensor but also the variation and deterioration of the engine.

On the other hand, the 2O 2 sensor system has the following unfavorable features in addition to these advantages.
Due to the O2 storage effect of the catalyst (a buffer-like action of storing or releasing oxygen), it takes a certain amount of time from the change of the air-fuel ratio on the upstream side of the catalyst to the change of the air-fuel ratio on the downstream side. As a result, the response of the downstream sensor is poor, and the control cycle becomes long. In addition, the control constant of the air-fuel ratio control is configured to be corrected based on the output of the downstream sensor, so that the correction speed of the control constant (to prevent the emission from being deteriorated by the overshoot caused by the poor response of the downstream sensor). It is necessary to slow down the control constant to change the air-fuel ratio) itself. As described above, the 2O2 sensor system has a disadvantage that it is necessary to lengthen the control cycle and slow the control speed.

Due to these disadvantages, the current 2O 2 sensor system is as follows. Since the control cycle is long and the control speed is slow, it is not suitable or impossible for the correction of the control constant in the transient state, and the control by the downstream sensor is performed only under steady conditions (for example, during constant speed running in medium to high speed). Feedback control for correction of constants (hereinafter referred to as "downstream O2 feedback")
Say. ) Could not be done.

Therefore, in a case other than the steady operation state, the downstream O2 feedback is stopped and the control constant (hereinafter, learned value) learned during the downstream O2 feedback is used. This learning is executed according to the timing when the output of the downstream sensor reverses between rich and lean. Specifically, the learning value is obtained by taking the average value of the control constant at the time of rich → lean inversion and the control constant at the time immediately before that of lean → rich inversion as the learning value, and at the time of lean → rich inversion, the control constant at that time and the immediately preceding The learning value was obtained by taking the average value of the control constants during rich-to-lean inversion. By adopting such a learning method, only the value at the time of inversion needs to be stored, so that the memory capacity can be small.

[0007]

However, since the conventional system is configured to learn only at the inversion timing of the downstream sensor, the learning is performed when the inversion of the output signal does not occur or is difficult to occur. There may be a problem that it is not performed or is hard to be performed, and the optimal control cannot always be reached.

For example, when the control constant at the time of starting the downstream O 2 feedback is far from the optimum value, inversion does not occur easily. And
If there is a shift change or acceleration / deceleration within the non-reversal period and the operating state becomes a transient state, the downstream O2 feedback will be stopped and the control constant will return to the initial value. It's gone.

Further, the reversing cycle of the downstream sensor in a general use state is considerably long, which is several tens to several hundreds of seconds. Therefore, not only when the initial value is far away from the optimum value, but also when driving in urban areas, the downstream O2 feedback often exits before the reversal occurs. Therefore, it becomes difficult to approach the optimum value due to the length of the control cycle and the slow control speed, which are features of the downstream sensor as described above.

As a result, in the conventional system, due to the characteristics of the downstream sensor, the exhaust emission is bad for a considerable period from the time of shipment from the vehicle, or it is difficult to return to the optimum exhaust emission performance after being initialized by repair or the like. There was a problem of doing.

Therefore, an object of the present invention is to provide an air-fuel ratio control device which can cover the weak points of such a 2O2 sensor system and can quickly converge to an optimum emission state in any state. To do.

[0012]

Therefore, an air-fuel ratio control system for an internal combustion engine according to the present invention is described in claim 1, and as shown in FIG. A catalyst, an oxygen sensor provided upstream and downstream of the three-way catalyst, and a control constant calculation that calculates a control constant necessary for air-fuel ratio control according to the output of the downstream sensor of the oxygen sensor. Means, learning means for learning the control constant calculated by the control constant calculation means at the timing when the output of the oxygen sensor on the downstream side reverses between rich and lean, and the output of the oxygen sensor on the upstream side. In the air-fuel ratio control device for an internal combustion engine, which comprises an air-fuel ratio control means for performing air-fuel ratio control based on the calculated or learned control constant, the output of the downstream oxygen sensor does not reverse. Even in the prescribed state Characterized by comprising also a forced learning means for performing the forced learning Once turned. To be more specific about this predetermined state, it can be said that the inversion does not occur even though it is predicted that the inversion occurs due to the original control characteristics of the oxygen sensor on the downstream side.

More specifically, it is described in claim 2, and FIG.
As exemplified in (a), in the air-fuel ratio control device for an internal combustion engine according to claim 1, the control constant calculated by the control constant calculating means is the predetermined state of the downstream oxygen sensor. It is possible to set the time when the output is calculated as a value that exceeds the control constant that was calculated at the time of the two-time inversion.

Further, as described in claim 3 and exemplified in FIG. 2 (b), in the air-fuel ratio control apparatus for an internal combustion engine according to claim 1 or claim 2,
It is possible to set a time when the control constant calculated by the control constant calculating means deviates from the current learning value by a predetermined amount or more.

Further, as described in claim 4 and illustrated in FIG. 2C, in the air-fuel ratio control device for an internal combustion engine according to any one of claims 1 to 3, the predetermined state is set. , It is possible to set the time when a predetermined time or more has elapsed since the previous output reversal. In the air-fuel ratio control device for an internal combustion engine according to the fourth aspect, as described in the fifth aspect, the predetermined time can be set in relation to the exhaust gas flow rate. Here, in relation to the exhaust gas flow rate,
It includes not only the meaning based on the exhaust gas flow rate itself, but also the case where it is determined based on other parameters closely related to the exhaust gas flow rate, such as the intake air amount, the intake pipe pressure, the engine speed, and the throttle opening.

[0016]

According to the air-fuel ratio control system for an internal combustion engine of the present invention,
While the feedback of the control constant based on the output signal of the oxygen sensor on the downstream side is being performed, the learning is basically performed at the inversion timing of the output signal.

On the other hand, even when such reversal does not occur easily, for example, when the control constant during feedback is calculated as a value exceeding the control constant calculated at the time of reversal two times before (for example, FIG. 2). If the calculated value of the control constant is increasing as shown in (a), the calculated value of the control constant exceeds the value before the previous calculation. If not, the calculated value of the control constant is decreased. When falling below)
If it becomes, learn (claim 2). Alternatively, when the control constant during feedback deviates from the current learning value by a predetermined amount or more, learning is performed even if inversion has not occurred (claim 3). Alternatively, if the inversion does not occur even after a lapse of a predetermined time from the previous output inversion, the learning is performed (claim 4). That is, in the air-fuel ratio control device of the present invention, learning can be performed in relation to the inversion of the output signal, and learning can be performed regardless of the inversion.

In the air-fuel ratio control apparatus according to the fifth aspect, the predetermined time is defined in relation to the exhaust gas flow rate based on the relationship between the exhaust gas flow rate and the time until the O2 storage performance of the catalyst is saturated. For example, when the exhaust gas flow rate is small, the O2 storage performance of the catalyst does not saturate easily. Therefore, if learning is performed before this saturation, the reliability of the learning value itself becomes extremely low. In order to eliminate such problems, the set time for forced learning is set longer when the exhaust gas flow rate is small, and conversely, when the exhaust gas amount is in a state where the catalyst O2 storage performance is saturated quickly, a short set time is set quickly. It is good to be able to learn.

[0019]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An air-fuel ratio control system for a vehicle internal combustion engine (hereinafter referred to as an engine) will be described below as an embodiment of the present invention. FIG. 3 shows its outline, and FIG. 4 shows the configuration of the control device.
The intake system 3 includes an air cleaner (not shown), a throttle valve 9, a surge tank 11, an air flow meter 13, a throttle position sensor 15, an intake air temperature sensor 17 and the like as various known structures. The throttle position sensor 15 incorporates an opening sensor 15a and an idle switch 15b. The idle switch 15b is a switch that is turned on when the engine 1 is idling.

The exhaust system 7 has a catalyst 118 and an upstream sensor 1.
9, a downstream sensor 119 is provided. Reference numerals 19 and 119 are electromotive force type sensors for detecting the oxygen concentration in the exhaust gas. In addition to the intake / exhaust system, igniter 21, distributor 2
3, a rotation speed sensor 25, a cylinder discrimination sensor 27, a cooling water temperature sensor 29, and the like. The rotation speed sensor 25 is the engine 1
A pulse is generated according to the rotational speed NE of. The cylinder block 1a of the engine 1 is cooled by circulating cooling water. The cooling water temperature is detected by a cooling water temperature sensor 29 provided in the cylinder block 1a.

The signals from the sensors such as 19, 119 and the rotation speed sensor 25 are input to the electronic control unit 30.
The electronic control unit 30 includes a CPU 31a, a ROM 31b, an RA
It is mainly composed of a microcomputer 31 having a built-in M31c and the like. The idle switch 15b, the rotation speed sensor 25, the cylinder discrimination sensor 27, the igniter 21, the heater energization control circuit 33, the drive circuit 35, etc. are connected to the input / output port of the microcomputer 31. A distributor 23 is connected to the igniter 21, and a spark plug 41 is connected to the distributor 23. The heater energization control circuit 33 uses the battery 37 as a power source, and the upstream sensor 1
The electric power supplied to the heater 19b of No. 9 is controlled. And
The heater 19b heats the detection element 19a.
Although not shown, the same structure as the upstream sensor is prepared for the downstream sensor and operates in the same manner. The drive circuit 26 is a circuit for driving the fuel injection valve 39.

A sensor for outputting analog signals from the air flow meter 13, the opening sensor 15a, the intake air temperature sensor 17, the cooling water temperature sensor 29, etc. to the input / output port of the microcomputer 31 via the A / D conversion circuit 42. Are connected. The A / D conversion circuit 42 further includes the output of the heater energization control circuit 33 for the upstream sensor and the current detection resistor 4
The terminal voltage of 3 and the output of the detection element 19a are input.
Similarly, the information for the downstream sensor is also input.

The electronic control unit 30 detects the operating state of the engine 1 based on the outputs of the various sensors including the oxygen sensor 19 and the output of the heater energization control circuit 33, and controls the operation of the engine 1. . The air-fuel ratio control will be described below. FIG. 5 shows a flowchart of a processing routine of air-fuel ratio feedback control (hereinafter referred to as "upstream O2 feedback") by an upstream sensor which is executed in the electronic control unit 30. This routine is based on the rich / lean judgment by the output voltage VU of the upstream sensor 19 and delay times TDR and TD which are predetermined control constants.
L, skip amount RSR, RSL, integration constants KIR, KI
This is a process for calculating the air-fuel ratio correction coefficient FAF in the air-fuel ratio feedback control using L, and the CPU 31a in the electronic control circuit 30 shown in FIG.
It is executed as a timer interrupt process for each. In addition, FA
The calculation result of F is reflected in the air-fuel ratio control by a known method.

In the execution of this process, first, it is judged whether or not the air-fuel ratio feedback execution condition is satisfied (S).
100). As the air-fuel ratio feedback execution condition, there are known conditions such as the water temperature level, the presence or absence of fuel cut, and whether or not the acceleration increase is being performed. If it is determined that the air-fuel ratio feedback execution condition is not satisfied (S1
00; NO), FAF = 1.0 is set (S103), and this routine is once ended.

When it is determined that the air-fuel ratio feedback execution condition is satisfied, the output voltage VU of the upstream sensor is fetched (step 105), and compared with the comparison voltage VRU, the actual air-fuel ratio rich / target air-fuel ratio rich / Lean is judged (step 107). When it is determined to be lean in this determination, the predetermined delay counter CDLY is counted down with 0 as an initial value (step 1
09-113), and when it becomes smaller than the minimum value TDL, the flag F1 is set to 0 (steps 115-11).
9). At this time, the delay counter CDLY also executes a process of guarding to the minimum value TDL (step 117). On the other hand, when it is determined to be rich in step 107, the delay counter CDLY is incremented from 0 (steps 121 to 125), and the delay counter CDL is compared with the maximum value TDR depending on the size.
The process of guarding Y to the maximum value TDR and setting the flag F1 to 1 is executed (steps 127 to 13).
1). That is, by the processing of steps 107 to 131,
When the air-fuel ratio is changed from rich to lean, the flag F1 is inverted from 1 to 0 with a delay of TDL, and when it is changed from lean to rich, the flag F1 is inverted from 0 to 1 with a delay of TDR. This delay is provided to further stabilize the rich / lean determination executed in the following process based on the inversion of the flag F1.

In the subsequent processing, it is first judged whether or not the flag F1 is inverted (step 133). If the flag is inverted, it is rich-lean inversion or lean-inversion.
It is judged whether or not F1 = 0 in order to know whether the rich is reversed (step 135). FA from rich to lean
The skip amount RSR is added to F (step 137), and if lean → rich, the skip amount RSL is subtracted from FAF (step 139). This causes FAF to be skipped. On the other hand, if the flag F1 is not inverted (step 133; NO), it is determined whether or not F1 = 0 in order to know whether it remains lean or rich (step 141). And if it stays lean, FA
The integration constant KIR is added to F (step 143), and if it remains rich, the integration constant KIL is subtracted from FAF (step 145). In this way, the value of FAF is determined by the processing of steps 133 to 145. In addition, FAF
In order to prevent the values from becoming too far apart, the guard processing is executed in the following steps 147 to 153 so that 0.8 <FAF <1.2.

In order to control the air-fuel ratio control center to be rich or lean, the skip amount RSR,
It goes without saying that the RSL and the integration constants KIR and KIL may be appropriately selected, but the delay time TDR,
The control center of the air-fuel ratio can also be finely adjusted to the rich side or the lean side by appropriately selecting TDL.

Next, the execution procedure of the feedback processing (hereinafter referred to as "downstream O2 feedback") for correcting the control constant used for upstream O2 feedback according to the value of the output voltage VD of the downstream sensor 119 will be described with reference to FIGS. It will be explained based on. In this downstream O2 feedback, as an example, the control constant for correction processing is set to the skip amount RSR, R.
It is SL.

This control constant correction processing is performed for a predetermined time longer than the upstream O2 feedback calculation cycle, for example, 524.
It is executed as a timer interrupt process every msec. This processing is performed by the output VD of the downstream sensor 119,
This is for finely adjusting the feedback control center so that the air-fuel ratio matches the catalyst window.

When this processing is executed, first, the execution condition of the downstream O2 feedback control is checked (step 201). Here, it is checked whether or not all of the following items are satisfied. Upstream O2 feedback control is in progress.

The water temperature is within a predetermined range (eg, 75 to 95 ° C.)
To be in. During steady operation (changes in rotation speed, intake air amount, throttle opening, etc. are small). The load (intake volume, boost, etc.) must be above a certain level.

The downstream sensor is active. This feedback condition is narrower than the upstream O2 feedback condition. For example, the downstream O2 feedback is executed only when the vehicle is traveling straight at a constant speed of 30 km / h or slow acceleration or slow deceleration. Configured. This is not specific to this embodiment,
The conditions are the same as those usually adopted in 2O2 sensor systems.

When it is determined in step 201 that the feedback control condition is not satisfied, that is, when the downstream O2 sensor output is open controlled (hereinafter referred to as downstream open), the learning value is stored in the backup memory. The mass production correction value RSPRO is added to the currently stored value RSRLRN, and this is set as the correction skip amount RSR '(step 203).

Here, RSPRO is for shifting the air-fuel ratio at the time of downstream opening with respect to the time of feedback, and its value is the voltage VPORT of the so-called mass-production correction port (not shown) by the search routine shown in FIG. The data is read (step 203a), and the characteristics shown in FIG. 10 are retrieved from the RSPRO map in a table (step 203b). As you can see from this map, RSPRO
The value of can be changed by the voltage applied to the mass production correction port, and even after masking the program of the microcomputer, the air-fuel ratio at the time of downstream opening can be finely adjusted. Further, with such a configuration, it is possible to finely adjust the appearance of the exhaust gas components based on the data such as the exhaust gas purification performance based on the actual running of the vehicles in the lot after the mass production is started.

After step 203, the routine proceeds to step 205 in FIG. 8, where adjustment is made from a map using the rotational speed NE and the intake air amount G / N per one rotation as a parameter (a table in which the characteristics of FIG. 11 are tabulated). Retrieve the quantity RSOFS. Then, PSOFS is added to the correction skip amount RSR 'to set as the skip amount RSR at the time of rich-to-lean inversion (step 207). After this, skip amount RS
RSR is subtracted from the total specified value RSSUM (for example, 10%) of R and RSL, and the skip amount R during lean-to-rich inversion
SL is calculated, and this routine is finished (step 20).
9). The RSOFS is provided to fill the difference between RSR convergence values (RSR values that end up due to the downstream O2 feedback control) that differ depending on the engine operating conditions, and to converge feedback quickly even when the conditions change.

When the feedback condition is satisfied in step 201, the output voltage VD of the downstream sensor is fetched and compared with the comparison voltage VRD to determine the air-fuel ratio as in steps 105 to 131 described above for the upstream O2 feedback control. Rich / lean determination is made, and the inversion process of the downstream O2 feedback flag F2 is executed (steps 211 to 241). During this period, the delay counter CDLY2 prepared for executing the downstream O2 feedback process is counted up / down while guarding with the maximum value TDR2 and the minimum value TDL2. The delay counter CDLY2 here also serves to make the rich / lean determination based on the inversion of the flag F2 more stable, and also has the maximum value TDR2 or the minimum value TD which is the delay time.
It is also possible to finely adjust the center of the air-fuel ratio feedback control to the rich side or the lean side by adjusting L2.

When the flag F2 inversion processing is completed in this way,
It proceeds to step 251 in FIG. 7 to check whether or not the flag F2 has been inverted, and at the time of inversion, proceeds to step 253 to execute the inversion process. Details of this inversion process are shown in FIG.
2 shows. In the inversion process, the current time TIMER is recorded as the inversion time TREV (step 301). Note that this processing is indispensable when the "non-inversion correspondence processing 2" described later is adopted, but it may not be executed when this is not adopted. In the following step 302, F2 =
It is determined whether it is 0 or not. When F2 = 0, that is, when rich → lean inversion, the correction skip amount RSR ′ calculated at this time is stored as RSRPL (step 305), and F2 = 1, that is, lean → rich inversion. If so, RSR 'at this point is stored as RSRPU (step 307). It should be noted that when the engine is started, RSRPL = RSRLRN−, for example, based on the learned value RSRLRN as RSRPL and RSRPU.
Initial values such as α and RSRPU = RSRLRN + β are set. After the process of step 305 or 307, the process proceeds to step 309, where RSRPL and RSRPU
Is calculated, and this is learned as RSRLRN. The learning value RSRLRN is backed up in the non-volatile memory and is retained even when the engine is stopped. Therefore, good exhaust gas characteristics can be immediately exhibited even after the engine is restarted.

When the reversal process (step 253) is completed in this way, the routine proceeds to step 254, and from the ΔRS map (a table of the characteristics of FIG. 13) using the intake amount as a parameter, a skip type for reversal is obtained. Incremental ΔRSP and Δ
Search RSM. Then, in order to determine whether the reversal was rich → lean or lean → rich, it is determined in step 255 whether or not F2 = 0. F2 = 0
If (rich → lean inversion), ΔRSP is added to RSR ′ that has been obtained so far (step 257),
After performing MAX guard processing in steps 259 and 261, the process proceeds to step 290 in FIG. If F2 = 1 (lean → rich inversion), ΔRSM is subtracted from RSR ′ (step 271), and M is calculated in steps 273 and 275.
After performing the IN guard process, the process proceeds to step 290.

On the other hand, if it is determined in step 251 that F2 is not inverted, the process proceeds to step 281.
In step 281, the integral increments ΔRSIP and ΔRSIM for non-inversion are retrieved from the ΔRS map corresponding to FIG. Then, in step 283, check F2,
When F2 = 0 (lean), ΔRSIP is added to RSR ′ (step 285), and when F2 = 1 (rich), ΔRSIP is added.
RSIM is subtracted from RSR '(step 287).
After these processes, the MAX guard process (step 25
9, 261) or MIN guard processing (step 27
3, 275) and then the process proceeds to step 290.

Here, ΔR in steps 254 and 281
The S map search will be described in more detail. As shown in FIG. 13, the ΔRS map is set so that the smaller the intake amount is, the larger each increment ΔRSP, ΔRSM, ΔRSIP, and ΔRSIM is. That is, the smaller the intake volume,
Control constant RSR 'by downstream O2 feedback control
Is easy to change. This is to improve the characteristic of the downstream O2 feedback that the control cycle becomes long when the intake air amount is low. The reason for having such characteristics is as follows.

The O 2 storage capacity of the catalyst is constant regardless of the engine operating conditions (strictly, it changes with temperature). On the other hand, the flow rate of exhaust gas discharged from the engine will change. Therefore, as shown in FIG. 14, when the O2 storage amount of the catalyst is 0, lean exhaust gas with the same air-fuel ratio (exhaust gas containing the same proportion of O2) was passed at a large flow rate and at a small flow rate. At times, when the flow rate is large, the O2 storage amount is saturated (that is, full). Therefore, if the same increment ΔRS is used regardless of the operating conditions of the engine, the control cycle becomes long at a small exhaust gas flow rate and becomes short at a large exhaust gas quantity. Therefore, the increment ΔRS is increased so that the change of the control constant RSR is increased at the time of the small exhaust gas flow rate, and the difference in the control cycle is eliminated. In this embodiment, instead of directly measuring the exhaust gas flow rate using the exhaust gas flow rate sensor, the intake air amount is used instead. This is because both are corresponding parameters. In addition to this, it is possible to substitute the intake pipe pressure, the engine speed, and the throttle opening instead of the intake air amount.

When the processing of steps 251 to 275 is completed in this way, next, the processing proceeds to step 290 of FIG. This process enables learning even during non-inversion, and in the present flow (FIGS. 6 to 8), it is configured to pass during inversion, but it may be passed only during non-inversion. The contents of the process of step 290 will be described with reference to FIG.

FIG. 15 shows the first non-reversal handling process.
The above method is referred to as non-inversion handling processing 1 below. In the non-reversal handling process 1, first, the RSR 'calculated at the present time is compared with RSRPL (the RSR' value at the previous rich-> lean reversal) (step 401). If RSR '<RSRPL, the current RSR' is set as a new RSRPL (step 403), and the learning value RSRLRN is updated using this new RSRPL (step 405). On the other hand, in step 401, RSR ′ ≧ R
If it is determined to be SRLRN, RSR 'is changed to RSRP.
It is compared with U (the RSR 'value at the last lean-> rich inversion) (step 407). And RSR '> RSRPU
If so, the current RSR 'is set as a new RSRPU (step 409), and the learning value RSRLRN is updated using this new RSRPU (step 405). In addition, R
If SRPL ≤ RSR '≤ RSRPU (step 40
If both 1 and 407 are NO), the learning value is not updated. That is, in the non-inversion handling process 1, first, when RSR 'becomes smaller or larger than RSR' in the previous two-time inversion, learning is performed despite the non-inversion. The above is the processing corresponding to the invention described in claim 2.

In the non-inversion correspondence processing 1, after this processing,
In step 411, the process corresponding to the invention described in claim 3 is also executed. In step 411, it is determined whether or not RSR 'at the current time point is smaller than the learned value RSRLRN at the current time point by a certain value KRSM. Small means that RSR 'is moving to a value that is considerably smaller than the learned value RSRLRN. At this time, RSRLRN = RSR '+ KRSM
Then, this process is terminated (step 413).
On the other hand, if it is determined that RSR 'is not smaller than RSRLRN-KRSM (step 41
1; NO), and RSR '> RSRLRN + KRS
It is determined whether or not it is P (step 415).
KRSP is a constant value different from KRSM. If it is judged to be large in this step 415, it means that RSR 'is a value considerably larger than the learning value. At this time, the processing is finished after updating RSRLRN = RSR'-KRSP (step 417).

Note that both steps 411 and 415 are NO.
If it is determined that RSR 'is the learning value RS at the present time.
Since it does not mean that the RLRN is out of the predetermined range, the present processing is ended without updating the learning value. This state means that RSR 'is near the learning value RSRLRN.

Thus, the non-reversal handling process (step 29)
When 0) is completed, the process proceeds to step 205 as shown in FIG. After that, the steps 205 to 209 are executed, and after performing the processing for calculating the skip amounts RSR, RSL for the upstream O2 feedback control from RSR 'almost in the same manner as at the time of the downstream opening described above, Exit the routine.

The above operation will now be described using a time chart. In FIG. 16, control constants for correction processing by upstream O2 feedback control and downstream O2 feedback will be described. Taking the case where the output VU of the upstream sensor 19 changes as shown in (a), the upstream sensor output VU is compared with the comparison voltage VRU, and the flag F1 changes as shown in (b). Here, the flag F1 changes with a delay of TDR for lean → rich and with a delay of TDL for rich → lean. FAF
Is controlled in a known manner in response to the change of the flag F1, and as shown in (c), RSR and RSL are skipped when F1 is inverted, and KIR and KIL are integrated when F1 is not inverted.

Next, the operation of the downstream O 2 feedback is shown in FIG.
This will be described using the time chart of No. 7. The downstream sensor output VD is processed with the same delay as the upstream sensor output VU and is reflected in the flag F2. The skip amount RSR used in the upstream O2 feedback and the base amount RSR 'for RSL calculation are controlled according to the change of F2. F2
Changes from 0 (lean) to 1 (rich), RSR '
The more skip-like increment ΔRSM is subtracted, and then it is reduced every 524 ms at a rate corresponding to the integral increment ΔRSIM. In addition, F2 is from 1 (rich) to 0 (lean)
Changes to, RSR 'is added with ΔRSP, and then ΔRSP
The rate of RSIP increases every 524 ms.

Here, if the state of F2 = 0 continues, time T
At 1, the value of RSR 'exceeds the value RSRPU of the previous lean-> rich reversal point (reversal point two times before). Therefore, the above-mentioned step 409 is executed and the time T
The value of RSR 'in 1 is newly set as RSRPU, the process proceeds to step 405, learning is performed, and the learned value is RSRL.
RN = (RSRPU + RSRPL) / 2 is updated.

After that, when F2 = 0 continues, the processes of steps 409 and 405 are repeatedly executed.
The learning value is updated. This is the operation corresponding to the invention of claim 2. However, since the change in RSRLRN during this period is due to the average of RSRPU and RSRPL,
It is half the rate of change of R '. Therefore, when this state continues and the time T2 shown in the figure is reached, RSR 'changes to RSRLRN.
Greater than + KRSP. Therefore, the determination in step 415 in the non-reversal handling process 1 is YES. As a result, the process of step 417 is executed, and thereafter, RSLRNRN = R until F2 is inverted.
The learning value SR'-KRSP will continue to be updated. This is the operation corresponding to the invention of claim 3.

Thus, according to this embodiment, the time T
From 1 onward, the learning value is updated based on a predetermined condition even if the flag F2 is not inverted. In this embodiment, if the condition is satisfied during the downstream feedback, the learning value is updated one by one. However, the learning value is used when the downstream O2 feedback condition is not satisfied and the downstream is opened. Of course, it is also possible to carry out all at once only at the time of transition. Further, in this embodiment, RS
Although the example in which R and RSL are learned even at the time of non-inversion is explained, the present invention may be applied to other control constants, for example, any one of the comparison voltage VRU, the delays TDR and TDL, and the integration constants KIL and KIR. Needless to say, the present invention may be applied to a plurality of types of control constants including RSR and RSL. This is because it is well known that the control air-fuel ratio can be changed by using each of these control constants alone or by combining a plurality of them, and therefore the present invention can be applied as it is.

Next, an embodiment corresponding to the invention described in claims 4 and 5 will be described. In this case, step 29
Only the contents of the non-inversion handling processing of 0 are different from the previous embodiment, but the others are the same. Therefore, only the non-inverting correspondence processing (hereinafter referred to as non-inverting correspondence processing 2) will be described.

FIG. 18 shows an execution procedure of the non-inversion handling process 2. In the non-reversal-time handling process 2, first, the comparison time TMREF is searched from a map (a table of the characteristic graph of FIG. 19) using the intake amount as a parameter (step 501). In this map, a value of control period when the downstream O2 feedback is converged × 1.2 to 2 is set. Then, (TIMER-TRE
V), that is, the elapsed time since the last reversal is the comparison time T
Check if it is longer than MREF (step 50)
3). If it is short here, this processing ends. On the other hand, if it is long, it is determined whether or not F2 = 0 (lean) (step 505). If lean, RSR 'at the present time is R
The SRPU is set and a temporary inversion point is set in preparation for the next learning (step 507). Further, a value obtained by subtracting a certain constant value KRSP from RSR 'is set as the learning value RSRLRN (step 509), and this processing is ended. On the other hand, if it is determined to be rich in step 505, R
Substituting RSR 'into SRPL (step 511), RS
R '+ KRSM is set as the learning value RSRLRN (step 513), and this processing ends. By the above-described processing, when the inversion of the downstream sensor does not occur for a long time, the learning can be forcibly performed based on the time. Since it is only necessary to set the optimum learning value at the time of downstream opening, the elapsed time from the previous inversion may be determined immediately before using the learning value, and the non-inversion handling processing 2 may be collectively performed there. Absent.

Next, a second embodiment will be described. This second embodiment is slightly different from the first embodiment described so far in terms of system, and uses so-called linearized PID control for upstream O2 feedback control. First, the contents of the upstream O2 feedback control, which is a feature of this second embodiment, will be described.

FIG. 20 shows an electronic control unit 30 according to the second embodiment.
5 is a flowchart of an upstream O2 feedback control processing routine executed by the above. Further, FIG. 21 shows a block diagram of feedback control realized by executing the routine of FIG. First, the processing routine of FIG. 20 will be described, and then the main air-fuel ratio feedback control will be described in detail according to the block diagram of FIG.

The upstream O 2 feedback control routine shown in FIG.
It is executed as a timer interrupt every 6 msec. In the execution, first, it is determined whether or not the air-fuel ratio feedback execution condition is satisfied (S600). Known air-fuel ratio feedback execution conditions include, for example, the water temperature level,
There are conditions such as the presence or absence of fuel cut and whether or not acceleration is being increased. If it is determined that the air-fuel ratio feedback execution condition is not satisfied (S600), this routine is once ended.

When it is determined that the air-fuel ratio feedback execution condition is satisfied (S600), the output voltage VU of the upstream sensor 19 is input (S610). Then, the standard excess air ratio λ1 is calculated from the output voltage VU (S620).
Here, the excess air ratio is a ratio of the actually supplied air amount when the air amount at the stoichiometric air-fuel ratio is used as a reference (= 1.0). The standard excess air ratio λ1 is determined by the upstream sensor 19
This is a value calculated by estimating the actual air amount in the air-fuel mixture from the oxygen concentration in the exhaust pipe based on the output voltage of. Next, it is determined whether the idle switch 15b is on (S630). When it is determined that the idle switch 15b is not on (S630), it is determined that the idle state is not established, and the characteristic graph for non-idle is referred to, and the step S
A control air excess ratio λ2 corresponding to the standard air excess ratio λ1 obtained in 620 is calculated (S640). Subsequently, the control excess air ratio λ2 is subtracted from the target excess air ratio λ0, and the subtracted value is set to the deviation Δλ (S650). Here, the target excess air ratio λ0 means an excess air ratio when the target air-fuel ratio is determined according to the running state of the vehicle, and for example, when the target air-fuel ratio is the theoretical air-fuel ratio, λ0 = 1.0. .

Next, it is determined whether or not the current operating condition is rapid acceleration (S660). When it is determined that the acceleration is not sudden (S
660), and retrieves the calculation parameter for PID control (S
670). On the other hand, when it is determined that the driving state is rapid acceleration (S
660), and a calculation parameter for PI control for acceleration is retrieved (S680).

The processing proceeds as described above in the non-idle state, but when it is determined in step S630 that the idle switch 15b is on, it is determined that the idle state is set, and the characteristic graph for the idle time is referred to to determine the standard excess air ratio. The control air excess ratio λ2 for λ1 is calculated (S690). Then, the control excess air ratio λ2 is subtracted from the target excess air ratio λ0, and the subtracted value is set to the deviation Δλ (S700).
Then, the PI control calculation parameter is searched (S71).
0).

From the calculation parameters (S670, 680, 710) retrieved according to the operating state such as the idle state, the non-idle state or the rapid acceleration state or the steady state in the non-idle state as described above, Then FAF is calculated (S720). Then, this routine is finished. F
The calculation result of AF is used for air-fuel ratio control by a known method, and the air-fuel ratio control based on FAF is executed.

Next, the upstream O2 feedback control realized by the execution of the processing routine of FIG. 20 will be described with reference to the equivalent block diagram of FIG. 21 (including downstream O2 feedback). Output voltage V of the upstream sensor 19
U is input to the linearizer 50. The linearizer 50 is realized by the steps S610 and 620. The linearizer 50 has the characteristic graph shown in FIG. 22 (the data of the characteristic graph is stored in the ROM 31b in advance). This characteristic graph shows the relationship between the output voltage VU of the upstream sensor 19 and the standard excess air ratio λ1. With reference to this characteristic graph, the linearizer 50 calculates the standard excess air ratio λ1 corresponding to the input output voltage VU.

The calculated standard excess air ratio λ1 is input to the non-idle correction linearizer 51 and the idle correction linearizer 53. The correction linearizer 51 is realized by the above step S640. The correction linearizer 53 is realized by the above step S690. The correction linearizer 51 is shown in FIG.
Or, having the characteristic graph for non-idle time shown in (B),
The correction linearizer 53 has a characteristic graph for idle time shown in FIG. 24 (data defining these is also stored in the ROM 31b in advance).

The non-idle characteristic graph of FIG. 23 and the idle characteristic graph of FIG. 24 both define the relationship between the standard excess air ratio λ1 and the control excess air ratio λ2. The characteristic graphs have some common basic relationships. This common basic relationship is shown in FIG.

That is, the basic relationship is the standard excess air ratio λ1.
= 1.0, that is, outside the range of 0.5% in total and 1% before and after centering on the value of the standard air excess ratio at the stoichiometric air-fuel ratio, regardless of the increase or decrease of the standard air excess ratio λ1. The relationship is that the control excess air ratio λ2 is maintained at a constant value without increasing or decreasing. Although it is also shown in FIG. 26 as the predetermined air-fuel ratio range, the variation in the output voltage VU of the upstream sensor 19 due to the individual difference of the upstream sensor 19 and the difference in the measured temperature is conspicuous mainly in the excess air ratio λ1 = 1.0. Before and after the above 0.5% each, it is outside the range of 1% width in total, and within the range of 1% width, the variation of the output voltage of the upstream sensor 19 is negligibly small. Has confirmed by experiments. Therefore, the output voltage of the upstream sensor 19 fluctuates in such a manner.
Outside the range of percent, the standard excess air ratio λ
Even if the control air excess ratio λ2 does not increase or decrease even if 1 increases or decreases, even if the output voltage of the upstream sensor 19 varies due to the individual difference or the difference in the measured temperature, the variation is controlled. The reason is that it is not reflected in λ2.

Next, a characteristic graph at the time of non-idling (see FIG.
3) and the characteristic graph at the time of idling (FIG. 24) will be described. As shown in FIGS. 23 (A) and 23 (B), in the characteristic graph at the time of non-idling, the standard air excess ratio λ1 is increased / decreased by deviating vertically or horizontally within the predetermined air-fuel ratio range of the 1% width. There is also defined a relationship in which the value of the control air excess ratio λ2 that increases / decreases accordingly is biased toward the rich side or the lean side as a whole. Here, rich or lean means that the air-fuel ratio is richer or leaner than the stoichiometric air-fuel ratio, and is shown by the symbols R and L in the figure (hereinafter the same).

On the other hand, in the characteristic graph at the time of idling, as shown in FIG. 24, the control air which increases / decreases in accordance with the increase / decrease of the standard air excess ratio λ1 in the predetermined air-fuel ratio range extending before and after the theoretical air-fuel ratio. The relationship is defined in which the increase / decrease rate of the excess rate λ2 is made lower than the basic increase / decrease rate (the increase / decrease rate of the chain line in the figure) centering on the theoretical air-fuel ratio.

The characteristic graph described above (see FIGS. 23 and 2).
4), the correction linearizer 51 and the correction linearizer 53 output the control excess air ratio λ2 corresponding to the standard excess air ratio λ1 input from the linearizer 50. The control excess air ratio λ2 output from the non-idle correction linearizer 51 is calculated by the deviation calculation circuit 55.
And the control excess air ratio λ2 output from the correction linearizer 53 for idle time is input to the deviation calculation circuit 57.

The deviations .DELTA..lamda. Between the control excess air ratio .lamda.2 and the target excess air ratio .lamda.0 are output from the respective deviation calculation circuits 55 and 57. After that, based on the deviation Δλ, the air-fuel ratio control described later is performed. In the air-fuel ratio control, the following control characteristics are basically obtained due to the difference in characteristics of the characteristic graph between the non-idle time and the idle time described above.

Both the non-idle time and the idle time are shown in FIG.
As shown in FIG. 5, when the standard excess air ratio λ1 out of the predetermined air-fuel ratio range is input, the control excess air ratio λ is constant.
2 is output. Therefore, when the standard air excess ratio λ1 is within the predetermined air-fuel ratio range, the control air excess ratio λ2 increases or decreases according to the increase or decrease of the standard air excess ratio λ1, but the standard air excess ratio λ1 does not exceed the predetermined air-fuel ratio range. If deviated, excess air ratio for control λ
2 will stop the increase and decrease. Even if the increase / decrease is stopped, the control air excess ratio λ2 has already been increased / decreased to a sufficiently large value or a sufficiently small value. The air-fuel ratio feedback control that is executed based on the deviation Δλ between the control excess air ratio λ2 and the target excess air ratio λ0 as described above is the difference between the control excess air ratio λ2 and the target excess air ratio λ0. The amount is reflected and the followability is good. As described above, the followability is excellent, but when the standard excess air ratio λ1 is outside the predetermined air-fuel ratio range, the increase or decrease of the control excess air ratio λ2 stops regardless of the increase or decrease of the standard excess air ratio λ1, The variation in the output of the upstream sensor 19 which becomes noticeable outside the predetermined air-fuel ratio range does not enter the air-fuel ratio feedback control. Therefore, there is no variation in control performance and no influence on emissions.

In the non-idle state, when the standard excess air ratio is within the predetermined air-fuel ratio range, the following control characteristics are basically realized. FIG. 27 shows an example of the characteristic graph at the time of non-idling. In this characteristic graph, two characteristics are shown by a solid line and a chain line. The solid line is the control excess air ratio λ2
It is a characteristic that the value of is biased toward the lean side as a whole. The chain line is a property without bias and is shown especially for comparison. When either of these two characteristics is registered in the correction linearizer 51, the control excess air ratio λ2 output by the correction linearizer 51 changes as shown in the time chart of FIG. 28. The change indicated by the solid line in the time chart is due to the characteristic indicated by the solid line in the characteristic graph (FIG. 27), and the change indicated by the chain line in the time chart is due to the characteristic indicated by the dashed line in the characteristic graph (FIG. 27).

In the change indicated by the chain line, the control excess air ratio λ
The average value of 2, that is, the position where the area when the control air excess ratio λ2 is on the rich side is the same as the area when it is on the lean side, overlaps with the theoretical air-fuel ratio. On the other hand, in the change indicated by the solid line, the average value of the control air excess ratio λ2, that is, the area when the control air excess ratio λ2 is on the rich side and the area when it is on the lean side (shown by diagonal lines) are The same value shifts to the lean side.

As a result of the shift to the lean side in this way, the air-fuel ratio control works to correct to the rich side. If the characteristic of FIG. 27 has a relationship in which the value of the control air excess ratio λ2 is biased to the rich side as a whole, the same action causes the air-fuel ratio control to correct to the lean side. That is, the center of the air-fuel ratio control can be finely adjusted by changing the setting of the deviation amount that biases the value of the control air excess ratio λ2 in the characteristic graph (FIG. 27) toward the lean side or the rich side. Therefore, even if the optimum air-fuel ratio at which the emission falls within the regulation value differs due to the individuality of the engine, the control to set the air-fuel ratio control center to this optimum air-fuel ratio can be easily realized by changing the setting of the above deviation amount. Will be.

In the second embodiment, the value of λ2 is used as the control target of the downstream O2 feedback, and the downstream sensor 119 is used.
The λ2 characteristic of the correction linearizer 51 is changed according to the output of Specifically, in FIG. 23A, λ1 = 1.0
The value of λ2 at that time is set as Kλ, and the characteristic of λ2 is changed. If the value of K.lamda. Is determined by the downstream O.sub.2 feedback, the value of .lamda.2 for .lamda.1 can be expressed by a well-known linear function formula or the like from the fixed values a and b and the value of K.lamda. In this way, the λ2 characteristic of the correction linearizer 51 can be changed.
The flowcharts are shown in FIGS. 29 to 34. Further, the map characteristics for controlling the Kλ value used in this flow by the downstream O2 feedback are shown in FIGS.
It should be noted that the comparison time TMREF used in the non-inversion handling processing 2
The same map as in the first embodiment is used as the search map. Kλ ′ is a calculated value corresponding to RSR ′,
KλPRO is a mass production correction value corresponding to RSPRO,
△ KλM, △ KλP, △ KλIM, △ KλIP are △ RS
M, ΔRSP, ΔRSIM, and ΔRSIP are increments, KλOFS is an adjustment amount corresponding to RSOFS, KMM and KPP are predetermined amounts corresponding to KRSM and KRSP, and KλPL and KλPU are RSPL and RSPU.
Is a calculated value at the time of inversion corresponding to, and KλLRN is RSR
It is a learning value corresponding to the LRN.

In the downstream O 2 feedback control processing in the second embodiment, as can be seen by comparing the flowcharts, the signs of addition and subtraction are slightly different from those in the first embodiment. This is because the air-fuel ratio changes in the rich direction when RSR 'is increased in the first embodiment, whereas it changes in the lean direction when Kλ is increased in the second embodiment. For this reason, the sign is opposite in consideration of the fact that Kλ decreases when lean and increases when rich. It is also different in that there is no process corresponding to step 209 of the first embodiment. The operation of the downstream O2 feedback in the other points has been described in detail in the first embodiment, and will be omitted here.

The upstream O 2 feedback control will be described in more detail below. First, the non-idle time will be described. The deviation Δλ output from the deviation calculation circuit 55 for non-idle time is output to the PID controller 59 for steady state and the PI controller 61 for rapid acceleration. PID
The controller 59 uses the transfer function G represented by the following mathematical expression 1.
The feedback control of c (S) is performed.

[0076]

[Equation 1]

However, Kp is a proportional constant, Ki is an integral constant, Kk is a differential constant, and Kd is a differential contribution constant. Differential element (Kk · S) / (1 + Kd · S)
Is an approximate derivative.

The calculation amount FAF is obtained in step S720 of the above-described processing routine (FIG. 20) in accordance with the calculation formula shown in Eq.

[0079]

[Equation 2]

However, FAF: air-fuel ratio correction coefficient, FAF
P: FAF proportional part, FAFI: FAF integral part, FA
FD: FAF differential part, Δλ: deviation Δλ. In addition,
For calculation, first calculate Δλ, then FAFP, FAF
I, FAFD, and finally FAF are performed every 16 msec. In addition, the term with i-1 in the formula is (16
It is a value calculated before msec).

On the other hand, PI controller 61 for sudden acceleration
Corresponds to the case where Kk = Kd = 0 in the equations 1 and 2. However, the values of Kp and Ki used may be different from those for steady-state use. Using these constants, the calculation amount FAF is obtained in step 720 of the above-described processing routine (FIG. 20).

As described above, the PID controller 59
And the calculation amount FA output from the PI controller 61
F is input to the first selection circuit 63. Further, the engine intake air amount change ΔG / N per time or rotation calculated from the output of the air flow meter 13 is input to the first selection circuit 63. This first selection circuit 63 has a processing routine (FIG. 20).
Of step S660. That is, the first selection circuit 63 determines whether it is the steady state or the rapid acceleration based on ΔG / N, and if it is the steady state, the PID controller 5
The calculation amount FAF from 9 is output to the downstream second selection circuit 67, and the calculation amount FAF from the PI controller 61 is output to the second selection circuit 67 during rapid acceleration.

Next, the idle time will be described. The deviation Δλ output from the deviation calculation circuit for idle time is output to the PI controller 65. The PI controller 65 is Kk = Kd = 0 in the equations 1 and 2 as in the case of the sudden acceleration.
Is equivalent to However, the values of Kp and Ki used are different from those for steady time and for rapid acceleration. Using these constants, step 720 of the processing routine described above (FIG. 20).
Then, the calculation amount FAF is obtained.

In this way, the PI controller 65 realizes PI control. The calculation amount FAF output from the PI controller 65 is input to the second selection circuit 67. The signal from the idle switch 15b is also input to the second selection circuit 67. The second selection circuit 67 performs the above step S6.
It is realized by 30.

The second selection circuit 67 determines whether the idle switch 15b is in the idle state or the non-idle state according to the contact state of the idle switch 15b.
9 or the calculation amount F output from the PI controller 61
The AF is output to the engine 1, and the calculation amount FAF output from the PI controller 65 is output to the engine 1 when idling. The engine 1 has a well-known configuration and the calculation amount FA
The air-fuel ratio feedback control is executed based on F.

As described above, in the second embodiment, since the linearized PID control is used for the upstream O2 feedback, the A / F control performance superior to that of the first embodiment is exhibited.
Further, the effect that the learning progresses even when the downstream sensor is not reversed is exactly the same as in the first embodiment.

Although some embodiments of the present invention have been described above, the present invention is not limited to these embodiments and can be implemented in various modes. For example, although the air flow meter is used in the engine control system of the embodiment, it goes without saying that the present invention can also be applied to a system using an intake pressure sensor and a throttle sensor.

[0088]

According to the air-fuel ratio control apparatus of the present invention, even if the output signal of the oxygen sensor on the downstream side does not reverse, the learning value is forcibly updated when it reaches a predetermined state. By covering the weak point of the 2O2 sensor system that the control cycle is long and the response is slow, the optimal emission state can be promptly converged regardless of the state.

The apparatus according to claim 2 effectively functions when the oxygen sensor on the downstream side does not easily invert from a certain point in time, and the apparatus according to claims 3 and 4 in such a case. Not only that, but it works effectively even in the case where no inversion occurs from the beginning. In particular, the apparatus according to claim 5 is suitable for maintaining the optimum emission characteristics without causing overshoot in consideration of the relationship with the O2 storage capacity of the catalyst.

[Brief description of drawings]

FIG. 1 is a schematic configuration diagram illustrating the present invention.

FIG. 2 is an explanatory diagram illustrating the invention according to claims 2 to 4 of the present invention.

FIG. 3 is a schematic configuration diagram of a system according to an embodiment.

FIG. 4 is a block diagram illustrating a configuration of a control device according to an embodiment.

FIG. 5 is a flowchart of upstream O2 feedback control.

FIG. 6 is a flowchart of downstream O 2 feedback control.

FIG. 7 is a flowchart of downstream O 2 feedback control.

FIG. 8 is a flowchart of downstream O 2 feedback control.

FIG. 9: RSP in downstream O 2 feedback control
It is a flow chart of RO search processing.

FIG. 10 is a graph showing characteristics of an RSPRO map used in the RSPRO search process.

FIG. 11: RS in downstream O 2 feedback control
It is a graph which shows the characteristic of the RSOFS map used by OFS search processing.

FIG. 12 is a flowchart of processing at the time of reversal in downstream O 2 feedback control.

FIG. 13 ΔR in downstream O 2 feedback control
9 is a graph showing the characteristics of the ΔRS map used in the S search process.

FIG. 14 is a graph showing the amount of O 2 storage in the catalyst and the output change of the downstream sensor in relation to the exhaust gas flow rate.

FIG. 15 is a flowchart of a non-inversion handling process 1 in the downstream O 2 feedback control.

FIG. 16 is a timing chart of upstream O 2 feedback.

FIG. 17 is a timing chart of the downstream O 2 feedback by the non-inversion handling processing 1.

FIG. 18 is a flowchart of a non-inversion handling process 2 in the downstream O 2 feedback control.

FIG. 19 is a comparison time TM used in the non-inverting correspondence processing 2.
It is a graph which shows the characteristic of the map for search of REF.

FIG. 20 is a flowchart of upstream O 2 feedback control in the second embodiment.

FIG. 21 is a block diagram of air-fuel ratio control realized in the second embodiment.

FIG. 22 is a characteristic graph that defines the relationship between the output voltage of the upstream sensor and the standard excess air ratio.

FIG. 23 is a characteristic graph that defines the relationship between the standard excess air ratio and the control excess air ratio when not idle.

FIG. 24 is a characteristic graph that defines the relationship between the standard excess air ratio and the control excess air ratio during idling.

FIG. 25 is a characteristic graph showing basic characteristics common to a non-idle characteristic and an idle characteristic.

FIG. 26 is a graph showing a variation caused by an individual difference or the like that occurs between the output of the oxygen sensor and the actual excess air ratio.

FIG. 27 is a characteristic graph showing one mode of the characteristic graph at the time of non-idling.

FIG. 28 is a time chart showing changes in the control excess air ratio.

FIG. 29 is a flowchart of downstream O 2 feedback control.

FIG. 30 is a flowchart of downstream O 2 feedback control.

FIG. 31: Kλ in downstream O 2 feedback control
It is a flow chart of PRO search processing.

FIG. 32 is a flow chart of processing during reversal in downstream O 2 feedback control.

FIG. 33 is a flowchart of non-inversion handling processing 1 in downstream O 2 feedback control.

FIG. 34 is a flowchart of a non-inversion handling process 2 in the downstream O 2 feedback control.

FIG. 35 is a graph showing the characteristics of a KλPRO map used in the KλPRO search process.

FIG. 36: Kλ in downstream O 2 feedback control
9 is a graph showing characteristics of a KλOFS map used in the OFS search process.

FIG. 37 ΔK in the downstream O 2 feedback control
6 is a graph showing characteristics of a ΔKλ map used in a λ search process.

[Explanation of symbols]

1 ... Engine, 13 ... Air flow meter, 15b
... Idle switch, 19 ... Upstream sensor, 25
... Rotation speed sensor, 30 ... Electronic control device, 31 ...
..Microcomputers, 118 ... Catalysts, 119
... Downstream sensor.

Claims (5)

[Claims] 1. A three-way catalyst provided in an exhaust pipe of an internal combustion engine, an oxygen sensor provided upstream and downstream of the three-way catalyst, and an output of a downstream sensor of the oxygen sensor. The control constant calculation means for calculating the control constant necessary for the air-fuel ratio control, and the control calculated by the control constant calculation means at the timing when the output of the oxygen sensor on the downstream side is inverted between rich and lean. Air-fuel ratio control of an internal combustion engine provided with learning means for learning a constant, air-fuel ratio control means for performing air-fuel ratio control based on the output of the upstream oxygen sensor and the calculated or learned control constant. In the device, even if the output of the oxygen sensor on the downstream side does not reverse,
An air-fuel ratio control apparatus for an internal combustion engine, further comprising a forced learning means for forcibly executing learning when a predetermined state is reached. 2. The air-fuel ratio control device for an internal combustion engine according to claim 1, wherein the control constant calculated by the control constant calculating means is set to a value before the output of the oxygen sensor on the downstream side before the predetermined state. An air-fuel ratio control device for an internal combustion engine, wherein a time calculated as a value exceeding a control constant calculated at the time of inversion is set. 3. The air-fuel ratio control device for an internal combustion engine according to claim 1 or 2, wherein the control constant calculated by the control constant calculating means as the predetermined state is relative to a current learning value. An air-fuel ratio control device for an internal combustion engine, wherein a time when a deviation from a predetermined value or more is set. 4. The air-fuel ratio control device for an internal combustion engine according to claim 1, wherein the predetermined state is set when a predetermined time or more has elapsed since the last output reversal. An air-fuel ratio control device for an internal combustion engine. 5. The air-fuel ratio control system for an internal combustion engine according to claim 4, wherein the predetermined time is determined in relation to an exhaust gas flow rate.