JPH09112310A - Air-fuel ratio controller for internal combustion engine - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve learning convergence while avoiding generation of convergence variation by increasing learning gain for correction if deviation more than a prescribed one is between a learned value and calculated value, in learning air-fuel ratio correction amount based on the detected output of an oxygen sensor on the downstream side of a catalyst. SOLUTION: According to the output of the first oxygen sensor on the upstream side, the first air-fuel ratio correction amount, or air-fuel ratio feedback correction coefficient α is set by proportional integral control. The second air-fuel ratio correction amount, or correction value PHOS is set according to the output of the second oxygen sensor on the downstream side. Learning gain is increased for correction by a prescribed period if deviation more than a prescribed value is between the learning value PHOSA and correction value PHOS.

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 device for an internal combustion engine, and more specifically, it has oxygen sensors on the upstream side and the downstream side of a catalyst, respectively, and controls the air-fuel ratio based on the detection results of these oxygen sensors. On the other hand, the present invention relates to an air-fuel ratio control device configured to learn an air-fuel ratio correction amount based on a detection result of an oxygen sensor provided on the downstream side.

[0002]

2. Description of the Related Art Conventionally, an oxygen sensor is provided on each of an upstream side and a downstream side of a three-way catalyst for exhaust gas purification provided in an exhaust passage, and an actual air-fuel ratio is calculated by using detection values of these oxygen sensors. An air-fuel ratio control device that controls to a target air-fuel ratio is known. For example, in the air-fuel ratio control device disclosed in Japanese Patent Laid-Open No. 4-72438, the air-fuel ratio feedback correction coefficient is set by proportional / integral control based on the detection result of the upstream oxygen sensor, while it is detected by the downstream oxygen sensor. Based on the rich / lean of the actual air-fuel ratio with respect to the target air-fuel ratio, by setting the correction value of the proportional operation amount in the proportional / integral control, the control point of the air-fuel ratio feedback control based on the detection result of the upstream oxygen sensor. It is configured to correct the deviation.

[0003]

By the way, in the case of learning the correction value of the proportional manipulated variable set based on the detection result of the downstream side oxygen sensor, if the learning gain is increased, the convergence becomes faster, but the convergence variation. However, the learning gain needs to be relatively small because of the problem that the value becomes large.

Therefore, when the learning value is cleared to the initial value, or when the correction request for the proportional manipulated variable changes significantly due to replacement of the oxygen sensor or the like, it takes time for the learning to converge, and during this time. There has been a problem that the desired accuracy of the air-fuel ratio control cannot be obtained. The present invention has been made in view of the above problems, and in the air-fuel ratio correction control based on the detection result of the oxygen sensor on the downstream side of the catalyst, improving convergence of learning while avoiding occurrence of convergence variation. With the goal.

[0005]

Therefore, an air-fuel ratio control system for an internal combustion engine according to the invention of claim 1 is constructed as shown in FIG. In FIG. 1, a first oxygen sensor and a second oxygen sensor are provided on the upstream side and the downstream side of an exhaust purification catalyst provided in an exhaust passage of an internal combustion engine, and their output values change in response to the oxygen concentration in exhaust gas. It is an oxygen sensor.

The first air-fuel ratio correction amount calculation means calculates the first air-fuel ratio correction amount according to the output value of the upstream first oxygen sensor. Further, the second air-fuel ratio correction amount calculation means is configured to change the second air-fuel ratio correction amount calculation means to the second
The air-fuel ratio correction amount is calculated. Further, the learning unit learns the second air-fuel ratio correction amount calculated by the second air-fuel ratio correction amount calculation unit.

The air-fuel ratio adjusting means adjusts the air-fuel ratio of the engine based on the learned values of the first and second air-fuel ratio correction amounts and the second air-fuel ratio correction amount. On the other hand, the learning gain increase correction means, when the learning value of the second air-fuel ratio correction value by the learning means and the second air-fuel ratio correction value calculated by the second air-fuel ratio correction amount calculation means have a deviation of a predetermined value or more. Then, the learning gain in the learning means is increased and corrected by a predetermined period.

According to this structure, the air-fuel ratio correction values are set based on the detection outputs of the first oxygen sensor and the second oxygen sensor, and the air-fuel ratio is adjusted based on these air-fuel ratio correction values. On the other hand, in learning the second air-fuel ratio correction amount calculated based on the detection output of the second oxygen sensor on the downstream side, the learned value and the value calculated based on the output of the second oxygen sensor If there is a deviation greater than a predetermined value, the learning gain is increased and corrected to ensure convergence responsiveness, but if the deviation is sufficiently small, the learning gain increase and correction are canceled and the convergence variation Occurrence is avoided.

According to the second aspect of the present invention, the learning gain increasing / correcting means increases / corrects the learning gain while the learning value is updated a predetermined number of times by the learning means. According to such a configuration, the learning gain can be increased and corrected by the number of times of update necessary and sufficient to bring the learned value close to the value calculated based on the output of the second oxygen sensor (second air-fuel ratio correction amount). .

According to the third aspect of the invention, the learning gain increase correction means has the learning value of the second air-fuel ratio correction value by the learning means and the second air-fuel ratio correction value calculated by the second air-fuel ratio correction amount calculation means. And, the learning gain in the learning means is increased and corrected for a predetermined period when the number of times that it is determined that there is a deviation of a predetermined value or more in the same direction becomes a predetermined number or more.

According to this structure, when the deviation temporarily shows a large value, the learning gain is not increased and corrected, and the learning value is proportional to the initial value or when the oxygen sensor is replaced. The learning gain is increased and corrected in a situation where the convergence is surely delayed unless the learning gain is increased and corrected, such as when the operation amount correction request changes significantly.

According to a fourth aspect of the present invention, the first air-fuel ratio correction amount calculation means makes the first air-fuel ratio so that the actual air-fuel ratio detected by the upstream first oxygen sensor approaches the target air-fuel ratio. The air-fuel ratio feedback correction coefficient as the fuel ratio correction amount is proportionally / integrally controlled, and the second air-fuel ratio correction amount calculation means sets the actual air-fuel ratio detected by the downstream second oxygen sensor to the target air-fuel ratio. The correction value of the proportional operation amount in the proportional control is calculated as the second air-fuel ratio correction amount so as to approach the fuel ratio.

According to this structure, the control point by the air-fuel ratio feedback correction coefficient calculated based on the output of the first oxygen sensor is corrected by the correction of the proportional operation amount based on the output of the second oxygen sensor, and the air-fuel ratio control is performed. While the accuracy is maintained, when learning the correction value of the proportional operation amount,
By switching the learning gain, it is possible to secure the convergence and suppress the convergence variation.

According to the present invention, the learning means averages the second air-fuel ratio correction amount every time the actual air-fuel ratio detected by the second oxygen sensor on the downstream side is reversed with respect to the target air-fuel ratio. A value is obtained, and the weighted average value of the average value and the learning value up to that point is updated and stored as a new learning value, and the learning gain increase correction means corrects the weighted weight in the weighted average as a learning gain. It was configured to do.

According to this structure, every time the air-fuel ratio detected by the second oxygen sensor is inverted, that is, the average value of the second air-fuel ratio correction amount is accurately obtained based on the peak value of the second air-fuel ratio correction amount. The weighted average value of the average value and the rare learning value is updated and stored as a new learning value, and the correction request by the second air-fuel ratio correction value is accurately learned. Here, the learning gain is increased and corrected and the convergence response is improved by correcting the weighted weight in the weighted average in the direction in which the weighting for the latest second air-fuel ratio correction value is made heavier.

[0016]

Embodiments of the present invention will be described below. In FIG. 2 showing the system configuration, the air filtered by the air cleaner 2 is sucked into the internal combustion engine 1 through the collector 5 and the intake manifold 6. The intake air flow rate of the engine 1 adjusted by the throttle valve 4 is measured by the air flow meter 3.

The intake manifold 6 is provided with an injector 7 for each cylinder, and the air-fuel mixture formed by the fuel injected from the injector 7 is sucked into the cylinder through the intake valve 8 and the spark plug 9 is introduced. Is ignited and burned by. In FIG. 2, 11 is a piston. The combustion exhaust gas is guided to a catalyst (exhaust gas purification catalyst) 14 via an exhaust valve 10 and an exhaust port 12, and the catalyst 14 emits CO, H
C and NOx are purified and discharged into the atmosphere.

First and second oxygen sensors 13 and 17 are provided on the upstream side and the downstream side of the catalyst 14, respectively. The first and second oxygen sensors 13 and 17 are sensors whose output values change in response to the oxygen concentration in the exhaust gas, and specifically, the oxygen concentration in the atmosphere as the reference gas and the oxygen concentration in the exhaust gas. It is an oxygen concentration cell type sensor that generates an electromotive force according to the ratio with the concentration.

The control unit 16 for controlling the fuel injection amount by the injector 7 and the ignition timing by the ignition plug 9 is supplied with the detection outputs from the air flow meter 3, the first and second oxygen sensors 13, 17. A detection signal from a water temperature sensor 15 that detects the cooling water temperature of the engine 1, a starter switch signal, a vehicle speed signal VSP, an engine rotation speed signal N, etc. are input.

Here, the control unit 16
Controls the fuel injection amount by the injector 7 according to the programs shown in the flow charts of FIGS. 3 to 6, thereby adjusting the air-fuel ratio of the engine intake air-fuel mixture.
In the present embodiment, the first air-fuel ratio correction amount calculation means, the second air-fuel ratio correction amount calculation means, the learning means, the air-fuel ratio adjustment means,
The function as the learning gain increasing / correcting means is as shown in the flow charts of FIGS.
16 are equipped with software.

The flowchart of FIG. 3 shows that the fuel injection amount T _I
Is shown. In S1, the intake air flow rate Q, the engine speed N, etc. detected by the air flow meter 3 are read. S
In 2, the basic fuel injection amount T _P is calculated as T _P = Q / N × K (K is a constant) based on the intake air flow rate Q and the engine rotation speed N.

In S3, the fuel injection amount T _I is set to T _I = T _P
It is calculated as × COEF × α + Ts. Here, T _P is the value calculated in S2, and COEF is the cooling water temperature T
Various correction coefficients set based on w, etc., α is an air-fuel ratio feedback correction coefficient, and Ts is a correction amount for correcting the invalid injection time by the battery voltage. The air-fuel ratio feedback correction coefficient α (first air-fuel ratio correction amount) is shown in FIG.
It is set by the proportional / integral control according to the flowchart of.

At S11, it is judged whether or not the air-fuel ratio feedback control condition is satisfied. If the control condition is not satisfied, the routine proceeds to S12, where the air-fuel ratio feedback correction coefficient α is set to 1.0 as the initial value. Close this program. On the other hand, when the control conditions are satisfied, the routine proceeds to S13, where the output OSR1 of the upstream first oxygen sensor 13 is set to A / D.
Convert and read.

In the next step S14, the output OSR1 is compared with a predetermined value SLF corresponding to the target air-fuel ratio (for example, the theoretical air-fuel ratio) to determine the rich lean of the actual air-fuel ratio with respect to the target air-fuel ratio. The output OSR1 is a predetermined value SLF
When it is less than the target air-fuel ratio, it is judged to be leaner than the target air-fuel ratio, the routine proceeds to S15, and 0 is set to the flag F ₁ .
On the other hand, when the output OSR1 exceeds the predetermined value SLF, it is determined that the output OSR1 is richer than the target air-fuel ratio, and S16
Proceed to and set 1 to the flag F ₁ .

In S17, it is determined whether or not the flag F ₁ is reversed by the present flag F ₁ setting, that is, whether the rich / lean state of the actual air-fuel ratio with respect to the target air-fuel ratio is reversed. When the air-fuel ratio is being reversed, the routine proceeds to S18, where it is determined whether or not the flag F ₁ is 0, in other words,
It is determined whether or not the state is reversed from rich to lean.

When reversing from rich to lean, S
Proceeding to 19, the correction value PHOS for correcting the proportional manipulated variables PL, PR in the proportional / integral control of the air-fuel ratio feedback correction coefficient α is set. The setting of the correction value PHOS will be described later. In the next step 20, the proportional operation amount PL and the correction value PHOS are added to the air-fuel ratio feedback correction coefficient α up to the previous time, and proportional control is performed in which the addition result is the latest correction coefficient α (α = α + PL +
PHOS).

On the other hand, when it is determined at S18 that the flag F ₁ is not 0, that is, when the lean-to-rich reverse operation is performed, the routine proceeds to S21, where the correction value PHOS is set, and then the routine proceeds to S22. Then, the correction coefficient α is proportionally controlled with α = α-PR + PHOS. Further, in S17, if it is determined not to be reversed when the air-fuel ratio, the process proceeds to S23, by determining the flag F _1, to determine the rich lean current air-fuel ratio.

When the flag F ₁ is 0 and the air-fuel ratio is lean, the routine proceeds to S24, where the integral operation amount IL is added to the correction coefficient α up to the previous time to update, and the flag F ₁ is 1 and the air-fuel ratio is empty. When the fuel ratio is rich, the routine proceeds to S25, where the integrated manipulated variable IR is subtracted from the correction coefficient α up to the previous time and updated. The flowchart of FIG. 5 shows how the correction value PHOS (second air-fuel ratio correction amount) is set in S19 and S21 of the flowchart of FIG.

In S31, it is determined whether or not it is in the air-fuel ratio control region by the second oxygen sensor 17, and if it is not in the control region, the process proceeds to S32, in which the learned value PHOSA learned in the control region is set as the correction value PHOS, The proportional operation amounts PL and PR are corrected based on the learned value PHOSA. The learning of the correction value PHOS will be described later.

On the other hand, when it is judged to be in the control region, the routine proceeds to S33, the output OSR2 of the second oxygen sensor 17 is compared with a predetermined value SLR, and the air-fuel ratio detected by the second oxygen sensor 17 is set to the target. It is determined whether the air-fuel ratio is rich or lean. Then, the output OSR2 is the predetermined value SL
When the air-fuel ratio is leaner than R and lean, the routine proceeds to S34, where the flag F ₂ is set to 0, while the output OSR2
Is above the predetermined value SLR and the air-fuel ratio is rich, the routine proceeds to S35, where the flag F ₂ is set to 1.

In S36, it is determined whether or not the air-fuel ratio detected by the second oxygen sensor 17 has been reversed, based on whether or not the flag F ₂ has been reversed. Then, when the air-fuel ratio is reversed, the routine proceeds to S37, where the flag F ₂ is discriminated. Flag F ₂
Is 0 and the air-fuel ratio is lean, the routine proceeds to S38, where the previous correction value PHOS is set to PHOS1. When it is determined in S38 that the air-fuel ratio is lean, it means that the air-fuel ratio is reversed from rich to lean.
In the air-fuel ratio rich state, the correction value PHOS is gradually decreased by the integral control as will be described later, so the PHOS1 is the peak value (minimum value) immediately before the correction value PHOS is controlled to increase. ) Will be shown.

In the next S39, the correction value PHOS up to the previous time is calculated.
Is added to the proportional operation amount PPHOSL, and proportional control is performed with the addition result as the current correction value PHOS. On the other hand, when it is determined at S37 that the flag F ₂ is not 0, that is, when the air-fuel ratio is reversed from lean to rich, the routine proceeds to S40,
As the peak value (maximum value) immediately before turning to the decrease control, the correction value PHOS up to the previous time is set in PHOS2.

In the next S41, the correction value PHOS up to the previous time is calculated.
A proportional operation amount PPHOSR is subtracted from the proportional operation amount PHOSR, and proportional control is performed by setting the addition result as the current correction value PHOS. And S
At 42, the correction value PHOS is learned. Such learning control will be described later. In S36, if it is determined not to be reversed when the flag F ₂ proceeds to S43, the flag F
Determine ₂

When the flag F ₂ is 0 and the air-fuel ratio is lean, the routine proceeds to S44, where the correction value PHO up to the previous time is set.
While adding the integral manipulated variable IPHOSL to S, the flag F
_{When 2} is 1 and the air-fuel ratio is rich, the routine proceeds to S45, where the integrated manipulated variable IP is calculated from the previous correction value PHOS.
The correction value PHOS is updated by subtracting HOSR. FIG.
The flowchart of FIG. 5 is S42 of the flowchart of FIG.
3 shows learning of the correction value PHOS in FIG.

In S51, the average value of the correction value PHOS and the current learning value PHOSA are compared. Specifically, the simple average value (PHOS) of the upper and lower peak values PHOS1 and PHOS2 set in S38 and S40 of the flowchart of FIG.
1 + PHOS2) / 2 minus learning value PHOSA, that is, (PHOS1 + PHOS2) / 2-PHOS
It is determined whether or not A is equal to or less than the negative predetermined value LPHOS.

The above-mentioned determination is to determine whether or not the learning value PHOSA is larger than the latest correction value PHOS by a predetermined value or more, and (PHOS1 + PHOS
2) / 2-PHOSA is equal to or less than the predetermined value LPHOS, the process proceeds to S52, and it is determined whether or not the counter CPHOS for counting the number of increase of the learning gain is zero.

The counter CPHOS is set to a predetermined value m when the learning gain is increased and corrected, and is set to be decremented by 1 each time the learning update is performed. When the learning gain is reduced to 0, the learning gain is normally set. It is supposed to return to the value. When the counter CPHOS is 0 and the learning gain increase control is not being performed, the process proceeds to S53 and the (P
HOS1 + PHOS2) / 2-PHOSA is a predetermined value LP
Counter CPHOS that counts that it was less than or equal to HOS
Increase L by 1.

When the counter CPHOS is not 0, S53 is shampooed and the process proceeds to S55. on the other hand,
If (PHOS1 + PHOS2) / 2-PHOSA is not equal to or less than the predetermined value LPHOS, the process proceeds to S54 and the counter CPHOSL is reset to zero. Therefore, the counter CPHOSL is (PHOS1 + PHOS2) /
2-PHOSA is counted up to a value exceeding 1 on the condition that the state in which 2-PHOSA is equal to or lower than the predetermined value LPHOS continues.

At S55, (PHOS1 + PHOS2) /
It is determined whether or not 2-PHOSA is greater than or equal to the positive predetermined value HPHOS, and thus it is determined whether or not the learning value PHOSA is smaller than the latest correction value PHOS by a predetermined value or more. Then, similarly to the above, (PHOS1 + P
HOS2) / 2-PHOSA is a positive predetermined value HPHO
If S or more, the condition is that the counter CPHOS is 0 (S56), (PHOS1 + PHOS2) / 2
-Count up the counter CPHOSH that counts the number of times PHOSA is determined to be greater than or equal to the plus predetermined value HPHOS (S57), (PHOS1 + PHOS
2) / 2-PHOSA is not greater than the predetermined plus value HPHOS, the counter CPHOSH is reset to zero (S58).

At S59, the counters CPHOSL, C
By determining whether or not one of PHOSH is greater than or equal to the predetermined value n, the learning value PHOSA and the correction value P
It is determined whether or not the number of times that the deviation from the HOS is equal to or greater than a predetermined value in the same direction is equal to or greater than a predetermined number. Then, the counters CPHOSL and CPHOSH
If either of the above is greater than or equal to the predetermined value n,
Go to S60, the counter CPHOSL, CPHOS
After resetting each H to zero, in S61, the correction value PHO
Weighting constant (weighting weight) G that determines the learning gain of S
As PHOS, GPHOS1 which is a value that makes the learning gain relatively large is set. Further, in S62, a predetermined value m is set in the counter CPHOS.

On the other hand, in S59, the counter CPHOS
When it is determined that both L and CPHOSH are less than the predetermined value n, the process proceeds to S63 and the counter CPHO is set.
It is determined whether S is 0 or not. If the counter CPHOS is 0, the process proceeds to S64, the weighting constant GPHOS is set to GPHOS2 which is a value that makes the learning gain relatively small, and the counter CPHO is set.
If S is not 0, the state in which GPHOS1 is set as the weighting constant GPHOS is held and S65 is set.
Proceed to.

At S65, the counter CPHOS is decremented by one. Here, the counter CPHOS has a predetermined value m.
During the period from the countdown to 0, the state in which GPHOS1 which is a value that makes the learning gain relatively large as the weighting constant GPHOS is set is held. In S66, the learning value PHOSA is updated and learned according to the following weighted average.

PHOSA = GPHOS × (PHOS1 +
PHOS2) / 2 + (1-GPHOS) × PHOSA In the above learning calculation formula, a weighting constant (weighting weight) G
The larger the PHOS (<1.0), the latest correction value P
The learning value PHOSA follows HOS quickly,
Since the learning gain is high, the weighting constants GPHOS1 and GPHOS2 are 1.0> GPHOS1
By setting in advance such that>GPHOS2> 0, when the deviation between the learning value PHOSA and the correction value PHOS is large, the deviation can be promptly eliminated.

For example, when the learning value is cleared to the initial value or when the oxygen sensors 13 and 17 are replaced, the correction value PHOS
When the correction request due to changes significantly, the learning value PH
The deviation between the OSA and the correction value PHOS becomes large, but the learning gain (weighting constant GPHOS) is set when the deviation is large due to the switching setting of the weighting constant GPHOS.
Is increased and corrected, the above deviation can be promptly eliminated, and thus the decrease in the air-fuel ratio control accuracy until the convergence can be suppressed to the minimum (see FIG. 7). On the other hand, the learning value P
When the deviation between the HOSA and the correction value PHOS becomes relatively small, the learning gain is returned to a relatively small value.
The occurrence of convergence variation can be suppressed (see FIG. 8).

In the above embodiment, the proportional operation amount in the proportional / integral control of the air-fuel ratio feedback correction coefficient α based on the output of the first oxygen sensor 13 is used as the second oxygen sensor.
Although the correction is performed based on the output of 17, the second air-fuel ratio correction amount calculated based on the output of the second oxygen sensor 17 is not limited to the correction value PHOS of the proportional operation amount. The delay time for proportional control of the correction coefficient α is set based on the output of the second oxygen sensor 17, and the delay time is learned, and the output OSR1 of the first oxygen sensor 13 is set.
Output OSR1 in rich / lean determination based on
The predetermined value SLF to be compared with may be corrected based on the output of the second oxygen sensor 17, and the correction value may be learned.

[0046]

As described above, according to the invention of claim 1, in learning the second air-fuel ratio correction amount calculated based on the detection output of the downstream second oxygen sensor,
The learning gain is increased and corrected when there is a deviation of a predetermined value or more between the learned value and the value calculated based on the output of the second oxygen sensor. Therefore, when the learning value is cleared to the initial value, When the request for correction of the proportional manipulated variable changes significantly due to replacement of the oxygen sensor, etc., the learning value can be converged at an early stage, while when the deviation is sufficiently small, the learning gain is made relatively small and converges. There is an effect that the occurrence of variations can be avoided.

According to the second aspect of the present invention, the learning gain can be increased and corrected only during a period necessary and sufficient to bring the learning value close to the value calculated based on the output of the second oxygen sensor. The effect is that you can do it. According to the invention as set forth in claim 3, it is necessary to increase and correct the learning gain as in the case where the learning value is cleared to the initial value or the request for correction of the proportional manipulated variable significantly changes due to replacement of the oxygen sensor or the like. There is an effect that the learning gain can be increased only in the situation described above.

According to the present invention, the control point by the air-fuel ratio feedback correction coefficient calculated based on the output of the first oxygen sensor is corrected by the correction of the proportional manipulated variable based on the output of the second oxygen sensor. Of course, there is an effect that the correction value learning of the proportional operation amount can be performed with good convergence and with suppressed convergence variation. According to the invention of claim 5, in a configuration in which a weighted average value of the average value of the second air-fuel ratio correction amount and a rare learning value is updated and stored as a new learning value, the weighted average corresponding to the learning gain is obtained. There is an effect that the learning value can be converged at an early stage by correcting the weighted weight in the direction so that the learning gain increases when the deviation between the learning value and the second air-fuel ratio correction value is large.

[Brief description of the drawings]

FIG. 1 is a configuration block diagram of the invention according to claim 1.

FIG. 2 is a system configuration diagram of the embodiment.

FIG. 3 is a flowchart showing how the fuel injection amount is calculated.

FIG. 4 is a flowchart showing a state of proportional / integral control of an air-fuel ratio feedback correction coefficient α (first air-fuel ratio correction amount).

FIG. 5 is a flowchart showing setting control of a correction value PHOS (second air-fuel ratio correction amount) of a proportional operation amount of a correction coefficient α.

FIG. 6 is a flowchart showing how the correction value PHOS is learned.

FIG. 7 is a diagram showing a correlation between learning gain and convergence.

FIG. 8 is a diagram showing a correlation between learning gain and convergence variation.

[Explanation of symbols]

 1 Internal Combustion Engine 3 Air Flow Meter 7 Injector 13 First Oxygen Sensor 14 Catalyst 16 Control Unit 17 Second Oxygen Sensor

Claims (5)

[Claims] 1. An exhaust purification catalyst provided in an exhaust passage of an internal combustion engine, and an exhaust purification catalyst which is provided upstream and downstream of the exhaust purification catalyst and whose output value changes in response to oxygen concentration in exhaust gas.
And a second oxygen sensor, first air-fuel ratio correction amount calculation means for calculating a first air-fuel ratio correction amount according to the output value of the upstream first oxygen sensor, and output of the downstream second oxygen sensor Second air-fuel ratio correction amount calculation means for calculating a second air-fuel ratio correction amount according to the value; learning means for learning the second air-fuel ratio correction amount calculated by the second air-fuel ratio correction amount calculation means; Air-fuel ratio adjustment means for adjusting the air-fuel ratio of the engine based on the learned values of the first and second air-fuel ratio correction amounts and the second air-fuel ratio correction amount; and a learned value of the second air-fuel ratio correction value by the learning means, Learning gain increase correction means for increasing and correcting the learning gain in the learning means when the second air-fuel ratio correction amount calculated by the second air-fuel ratio correction amount calculation means has a deviation of a predetermined value or more; Air-fuel ratio control of an internal combustion engine characterized by including Apparatus. 2. The air-fuel ratio control of an internal combustion engine according to claim 1, wherein the learning gain increase correction means increases and corrects the learning gain while the learning value is updated a predetermined number of times by the learning means. apparatus. 3. The learning gain increase correction means, the learning value of the second air-fuel ratio correction value by the learning means, and the second air-fuel ratio correction value calculated by the second air-fuel ratio correction amount calculation means,
The learning gain in the learning means is increased and corrected by a predetermined period when the number of times it is determined that the deviation in the same direction is greater than or equal to a predetermined value is greater than or equal to a predetermined number. Air-fuel ratio controller for internal combustion engine. 4. The first air-fuel ratio correction amount calculating means calculates the air-fuel ratio as the first air-fuel ratio correction amount so that the actual air-fuel ratio detected by the upstream first oxygen sensor approaches a target air-fuel ratio. It is configured to perform proportional / integral control of the fuel ratio feedback correction coefficient, and the second air-fuel ratio correction amount calculation means makes the actual air-fuel ratio detected by the second oxygen sensor on the downstream side close to the target air-fuel ratio. The air-fuel ratio control device for an internal combustion engine according to any one of claims 1 to 3, wherein a correction value of a proportional operation amount in the proportional control is calculated as a second air-fuel ratio correction amount. 5. The learning means obtains an average value of the second air-fuel ratio correction amount every time the actual air-fuel ratio detected by the second oxygen sensor on the downstream side is reversed with respect to the target air-fuel ratio, The weighted average value of the average value and the learning value up to that point is updated and stored as a new learning value, and the learning gain increase correction means corrects the weighted weight in the weighted average as a learning gain. The air-fuel ratio control device for an internal combustion engine according to claim 4.