JPH0942025A - Air/fuel ratio controller of internal combustion engine - Google Patents

Abstract

PROBLEM TO BE SOLVED: To increase a change of renewing a learnt value properly so that it may quickly respond to the state of an engine every time with high accuracy by learning an air/fuel ratio correction amount matching the operational state of the engine, while changing the amount of renewal in response to intake frequency or intake time depending upon an air/fuel ratio correction factor. SOLUTION: An air/fuel ratio controller is provided with a fuel injection valve 10 for making injection fuel supply to the combustion chamber 3 of an internal combustion engine 1, and a means 11 for detecting the air/fuel ratio of the engine from its exhaust gas. It is also provided with an electronic controller 30 so as to carry out both air/fuel ratio feed-back control and air/fuel ratio learning control. A feed-back control means makes its feed-back control of the operational quantity of the fuel injection valve 10 depending upon an air/fuel ratio correction factor so computed as to converge the air/fuel ratio of the engine in a specified range. A learning control means learns the air/fuel ratio correction quantity corresponding to the operational state of the engine while changing the amount of renewal according to intake frequency or intake time of the air/fuel ratio correction factor depending upon the air/fuel ratio correction factor itself. Thereby, the chance of renewing a learnt value can be increased properly and certainly.

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 mounted on an automobile or the like, and more particularly to a device that uses learning control in combination with the air-fuel ratio control to improve the accuracy of air-fuel ratio control based on the learning. The present invention relates to improvement of a learning control structure for further improvement.

[0002]

2. Description of the Related Art Conventionally, as an air-fuel ratio control device which also uses such learning control, for example, Japanese Patent Publication No. 62-12382.
The device described in Japanese Patent Laid-Open No. 61-89946 and the like are known.

Incidentally, in the device disclosed in Japanese Patent Publication No. 62-12382, the learning control method of the air-fuel ratio is as follows: When the air-fuel ratio is changed to rich / lean by an air-fuel ratio sensor provided in the exhaust system of the engine, Alternatively, at the time when the proportional-plus-integral correction direction changes, a predetermined number of engine state correction amounts as learning values are taken in, and the arithmetic mean value of these correction amounts is calculated. -Based on the calculated arithmetic mean value, the engine state correction amount as the learning value is corrected. Such as,
The air-fuel ratio of the engine is feedback-controlled to the target air-fuel ratio according to the engine state correction amount thus corrected, that is, the learned value.

According to such a learning control method, the center of the air-fuel ratio can be clarified without being affected by the fluctuation cycle of the air-fuel ratio, and therefore the accuracy based on the highly accurate engine state correction amount (learning value) can be obtained. It becomes possible to perform high air-fuel ratio control.

Further, in the device disclosed in Japanese Patent Laid-Open No. 61-89946, in order to obtain the engine state correction amount as a learning value, the following is performed: -The air-fuel ratio of the engine is feedback-controlled based on the output of the air-fuel ratio sensor. The average value of the air-fuel ratio correction coefficient found in the system within a predetermined period is calculated. -The deviation between the calculated average value of the air-fuel ratio correction coefficient and the theoretical value (for example, "1.0") is further calculated. The amount of increase / decrease (correction amount) of the learning value is changed according to the calculated deviation amount. Such as learning control structure,
The air-fuel ratio is corrected and controlled according to a learning value in which an increase / decrease amount is variably set according to the deviation amount.

As described above, if the increase / decrease amount of the learning value is changed according to the deviation amount between the average value of the air-fuel ratio correction coefficient and the theoretical value, the followability as the learning control system is greatly improved. become.

[0007]

By using such learning control together with the air-fuel ratio control of the internal combustion engine, the aging, secular change, etc. of those engines are suitably corrected and the accuracy of the air-fuel ratio control is also kept extremely high. Become so.

However, in any of the conventional learning control structures described above, in order for the learning value to be updated, it is necessary to take in the learning elements such as the air-fuel ratio correction coefficient to a predetermined number. That is, if there is a case where the learning condition is deviated from before reaching the predetermined number of learning elements, the learning value will not be updated forever. If such a state is prolonged, the control state of the air-fuel ratio itself may not correspond to the engine state at each time.

The present invention has been made in view of the above circumstances, and appropriately increases the chances of updating the learning value to realize more accurate air-fuel ratio control that immediately responds to the engine state at each time. An object of the present invention is to provide an air-fuel ratio control device for an internal combustion engine that can perform the above.

[0010]

In order to achieve such an object, in the invention as set forth in claim 1, a fuel injection valve for injecting and supplying fuel pressure-fed from a fuel tank to a combustion chamber of an internal combustion engine, and an exhaust gas of the engine. An air-fuel ratio detection means for detecting the air-fuel ratio of the engine from the exhaust gas provided in the system, an air-fuel ratio correction coefficient calculation means for calculating an air-fuel ratio correction coefficient according to the detected air-fuel ratio, and the engine Based on the calculated air-fuel ratio correction coefficient, feedback control means for feedback-controlling the operation amount of the fuel injection valve based on the calculated air-fuel ratio correction coefficient so that the air-fuel ratio converges within a predetermined range. , Learning control means for learning the air-fuel ratio correction amount according to the operating state of the engine while changing the update amount according to the number of times of intake or the intake time, and the learned air-fuel ratio correction amount. Depending to constitute the air-fuel ratio control apparatus comprises a correction means for correcting the amount of operation of the fuel injection valve which is the feedback control.

Further, in the invention of claim 2, in the configuration of the invention of claim 1, the learning control means includes an average value calculation means for calculating an average value of the fetched air-fuel ratio correction coefficients, and A deviation amount calculation means for calculating the deviation amount between the calculated average value and the theoretical value, and an update amount of the air-fuel ratio correction amount according to the operating state of the engine which is a learning value according to the calculated deviation amount. Update amount calculation means for calculating, and update amount correction means for correcting the calculated update amount in accordance with the number of times or time of taking in the air-fuel ratio correction coefficient,
A learning value calculating means for updating the learning value based on the corrected update amount is provided.

Further, in the invention according to claim 3, in the configuration of the invention according to claim 1, the learning control means is a learning condition judging means for judging whether or not a learning condition is satisfied, and the learning condition is satisfied. On the condition that the air-fuel ratio correction coefficient is taken in, the average value calculating means for calculating the average value of the taken-in air-fuel ratio correction coefficient, and the number of times of taking in the air-fuel ratio correction coefficient or the taking-in time reaches a predetermined number of times or a predetermined time. A first deviation amount calculation means for calculating a first deviation amount between the calculated average value and the theoretical value, and the calculated first deviation amount calculation means.
A first update amount calculation means for calculating a first update amount of the air-fuel ratio correction amount according to the operating state of the engine, which is a learning value, according to the deviation amount of, and the calculated first update amount. First learning value calculation means for updating a learning value based on the second learning amount, and a second deviation amount between the theoretical value and the average value calculated immediately before the air-fuel ratio correction coefficient immediately after the learning condition is not satisfied. A second deviation amount calculation means, and a second update for calculating a second update amount of the air-fuel ratio correction amount corresponding to the operating state of the engine, which is a learning value according to the calculated second deviation amount. An amount calculation means, an update amount correction means for correcting the calculated second update amount in accordance with the number of times of intake or an intake time of the air-fuel ratio correction coefficient, and the corrected second update amount. A second learning value calculating means for updating the learning value based on To.

Further, in the invention according to claim 4, the fuel injection valve for injecting and supplying the fuel pressure-fed from the fuel tank to the combustion chamber of the internal combustion engine and the exhaust system of the engine are provided, and the exhaust gas from the engine is used. Air-fuel ratio detecting means for detecting the air-fuel ratio, air-fuel ratio correction coefficient calculating means for calculating an air-fuel ratio correction coefficient according to the detected air-fuel ratio, and the air-fuel ratio of the engine is converged within a predetermined range. As described above, the feedback control means for feedback-controlling the operation amount of the fuel injection valve based on the calculated air-fuel ratio correction coefficient, and the number of times of intake or the intake time are changed according to the calculated air-fuel ratio correction coefficient, Learning control means for learning an air-fuel ratio correction amount according to the operating state of the engine, and operation of the fuel injection valve that is feedback-controlled according to the learned air-fuel ratio correction amount. It comprises a correction means for correcting an amount constituting the air-fuel ratio control apparatus for an internal combustion engine.

According to a fifth aspect of the present invention, in the configuration of the fourth aspect of the invention, the learning control means includes an average value calculation means for calculating an average value of the fetched air-fuel ratio correction coefficients. Deviation amount calculation means for calculating the deviation amount between the calculated average value and the theoretical value, and a threshold value setting for setting the threshold value for the number of times of intake of the air-fuel ratio correction coefficient or the intake time according to the calculated amount of deviation. Means and
Learning to update the air-fuel ratio correction amount according to the operating state of the engine, which is a learning value, based on a predetermined update amount, provided that the number of times the air-fuel ratio correction coefficient is taken in or the time taken up reaches this set threshold value. And a value calculation means.

Further, in the invention according to claim 6, in the structure of the invention according to claim 5, the air-fuel ratio correction amount according to the operating state of the engine, which is the learning value according to the calculated deviation amount. Further, the learning value calculating means is configured to update the learning value based on the calculated update amount.

[0016]

In general, in learning control of the air-fuel ratio, in order to update the air-fuel ratio correction amount according to the operating state of the engine, which is the learned value, the number of times learning elements such as the air-fuel ratio correction coefficient are taken Alternatively, it is necessary that the acquisition time reaches a predetermined number of times or a predetermined time. If there is a case where the learning condition is deviated from the learning condition before the number of times the learning element is captured or the capture time reaches a predetermined number of times or a predetermined time,
As mentioned above, the learning value will not be updated over time.

Therefore, in the invention according to claim 1, as the learning control means, the operating state of the engine is changed based on the air-fuel ratio correction coefficient while changing the update amount according to the number of times of intake or the time of intake. The air-fuel ratio correction amount according to is learned. A means having such a learning control structure is adopted.

According to such a learning control structure, the number of times the air-fuel ratio correction coefficient, which is a learning element, is taken in or the time taken in does not reach a predetermined number of times or a predetermined time, The learning value is updated by the amount that That is, the chances of updating the learned value are surely increased, and more accurate air-fuel ratio control that immediately responds to the engine state at each time is realized.

According to the second aspect of the invention,
The learning control means is an average value calculation means for calculating an average value of the fetched air-fuel ratio correction coefficients. A deviation amount calculation means for calculating the deviation amount between the calculated average value and the theoretical value. Update amount calculation means for calculating the update amount of the air-fuel ratio correction amount corresponding to the operating state of the engine, which is a learning value, according to the calculated deviation amount. Update amount correction means for correcting the calculated update amount according to the number of times the air-fuel ratio correction coefficient is taken or the time taken. -Learning value calculation means for updating the learning value based on the corrected update amount. Can be configured as each.

Basically, it is possible to set the update amount of the learning value according to only the number of times the air-fuel ratio correction coefficient is taken in or the time taken. However, as in the invention according to claim 2, the basic update amount of the learning value is calculated according to the deviation amount between the average value of the air-fuel ratio correction coefficient which is the learning element and the theoretical value, and the calculated basic update amount is calculated. Is corrected according to the number of times the air-fuel ratio correction coefficient is taken in or the time taken in, the updated learning value can also be obtained as a more accurate value.

According to the invention of claim 3,
The learning control means is a learning condition determination means for determining whether or not a learning condition is satisfied. An average value calculating means for calculating the average value of the taken-in air-fuel ratio correction coefficients, provided that this learning condition is satisfied. A first deviation amount calculation for calculating a first deviation amount between the calculated average value and the theoretical value on condition that the number of times of taking in the air-fuel ratio correction coefficient or the taking-in time reaches a predetermined number of times or a predetermined time means. First update amount calculation means for calculating a first update amount of the air-fuel ratio correction amount according to the operating state of the engine, which is a learned value according to the calculated first deviation amount. First learning value calculation means for updating the learning value based on the calculated first update amount. Second deviation amount calculation means for calculating a second deviation amount between the theoretical value and the average value calculated immediately before the air-fuel ratio correction coefficient immediately after the learning condition is not satisfied. Second update amount calculation means for calculating a second update amount of the air-fuel ratio correction amount according to the operating state of the engine, which is a learning value, according to the calculated second deviation amount. Update amount correction means for correcting the calculated second update amount according to the number of times or time of taking in the air-fuel ratio correction coefficient at that time. Second learning value calculation means for updating the learning value based on the corrected second update amount. It is also possible to configure each as.

Here, each of the first means constitutes an ordinary learning control system for updating the learning value on the condition that the learning condition is satisfied, and each of the second means is constituted by the second means. A learning control system is configured to update the learning value in the form according to the invention of claim 2 only immediately after the learning condition is changed from being satisfied to being not satisfied.

According to this structure, even if the learning condition is deviated from the learning condition before the air-fuel ratio correction coefficient, which is a learning element, is fetched a predetermined number of times or for a predetermined time, the highly accurate learning described in the invention according to claim 2 is performed. It is also executed immediately after the learning condition is deviated.

Therefore, in this case as well, the chances of updating the learning value are surely increased, and more accurate air-fuel ratio control that responds immediately to the engine state at each time is realized.

By the way, in view of a normal learning control structure in which the learning value is updated under the condition that the learning number such as the air-fuel ratio correction coefficient or the like is taken in or the taking time reaches a predetermined number of times or a predetermined time, the number of taking-in times is increased. Alternatively, even if the acquisition time is properly shortened, the chances of updating the learning value also increase.

That is, as the invention according to claim 4, as the learning control means, the operating state of the engine is changed while changing the number of times of intake or the time of intake according to the calculated air-fuel ratio correction coefficient. The corresponding air-fuel ratio correction amount is learned. Even by adopting such a learning control structure, as long as the change in the number of times the air-fuel ratio correction coefficient is taken or the change in the time taken is shortened, the opportunity for updating the learning value is surely increased. Therefore, also in this case, more accurate air-fuel ratio control that immediately responds to the engine state at each time can be realized.

According to the invention of claim 5,
Such learning control means are: average value calculation means for calculating an average value of the taken-in air-fuel ratio correction coefficient. A deviation amount calculation means for calculating the deviation amount between the calculated average value and the theoretical value. Threshold setting means for setting a threshold for the number of times the air-fuel ratio correction coefficient is loaded or the loading time according to the calculated deviation amount. Update the air-fuel ratio correction amount corresponding to the operating state of the engine, which is a learning value, based on a predetermined update amount, on condition that the number of times the air-fuel ratio correction coefficient is taken in or the time taken up reaches the set threshold value. Learning value calculation means. Can be configured as each.

As described above, if the threshold value for the number of times the air-fuel ratio correction coefficient is taken or the time taken for the air-fuel ratio correction coefficient is set according to the amount of deviation between the average value and the theoretical value of the air-fuel ratio correction coefficient, which is a learning factor, The set threshold value also becomes a more appropriate value according to the operating state of the engine in each case.

Here, the case where the deviation amount is large means the case where the air-fuel ratio correction coefficient largely deviates from the theoretical value. At this time, the threshold value is set so as to correct the air-fuel ratio to an appropriate value at an early stage. Set to a relatively small value.

On the other hand, the case where the deviation amount is small means that the deviation between the air-fuel ratio correction coefficient and the theoretical value is small,
At this time, the threshold value is set to a relatively large value in order to correct the air-fuel ratio more accurately.

It is assumed that the maximum value of the threshold value is set to the same level as the threshold value adopted in the above-mentioned conventional learning control structure. As a result, the opportunity for updating the learning value will not be reduced as compared with the conventional learning control structure.

According to the invention of claim 6, an update amount for calculating the update amount of the air-fuel ratio correction amount corresponding to the operating state of the engine, which is the learning value, according to the calculated deviation amount. Computing means. If the learning value calculation means is configured to update the learning value based on the calculated update amount, the updated learning value can be obtained as a value with higher accuracy. .

[0033]

【Example】

(First Embodiment) FIG. 1 shows a first embodiment of an air-fuel ratio control system for an internal combustion engine according to the present invention.

The device of this embodiment is a device that uses learning control in combination with air-fuel ratio control of an internal combustion engine mounted on an automobile, etc., and the learning value update amount is changed according to the number of times of taking in the air-fuel ratio correction coefficient. Is configured as a device having a control structure.

First, the configurations of the multi-cylinder internal combustion engine 1 and its peripheral devices to which the apparatus of this embodiment is applied will be described with reference to FIG. The internal combustion engine 1 is roughly divided into an intake system 2, a combustion chamber 3, and an exhaust system 4. Of these, the intake system 2 and the exhaust system 4 will be briefly described below.

In the engine 1, the intake system 2 is composed of an air cleaner (not shown), a throttle valve 5 and a surge tank 6 from the upstream side thereof, and an intake pressure sensor 7 is provided at each part thereof. A throttle position sensor 8 and an intake air temperature sensor 9 are provided respectively.

Of these sensors, the intake pressure sensor 7 is
The throttle position sensor 8 is a sensor arranged in the surge tank 6 to detect the intake pipe negative pressure PM, and the throttle position sensor 8 is a sensor arranged in the vicinity of the throttle valve 5 to detect its opening. The throttle position sensor 8
Includes an opening sensor 8a that outputs opening information of the throttle valve 5 and an idle switch 8b that is turned on when the engine 1 is idling. The intake air temperature sensor 9 is a sensor that detects the temperature of the air taken into the engine 1.

The intake system 2 includes a fuel injection valve 10
Is provided. The fuel pressure-fed from a fuel tank (not shown) is injected and supplied into the engine 1 according to the operation of the fuel injection valve 10, and is mixed with the air taken in through the intake system 2.

On the other hand, the exhaust system 4 has an oxygen sensor (O 2 sensor) 11 for detecting the oxygen concentration of the gas discharged from the combustion chamber 3. This oxygen sensor 11
It is an electromotive force type sensor, and as information indicating the air-fuel ratio of the engine 1, that is, the ratio of the mixed air and fuel,
A signal indicating "rich (R)" or "lean (L)" is output.

In addition, the engine 1 is provided with an igniter 12 which is an ignition device and a distributor 13 which is a distributor, and the distributed ignition voltage is applied to an ignition plug 14 provided in the combustion chamber 3 of each cylinder. Is applied.

Further, the distributor 13 is provided with a rotation speed sensor 15 and a cylinder discrimination sensor 16, through which the rotation speed NE of the engine 1 is detected and the combustion cylinder is discriminated.

The cylinder block 1a of the engine 1 is cooled by the circulating cooling water, and the temperature of the cooling water is detected by the water temperature sensor 17 provided in the cylinder block 1a. Become so.

In the internal combustion engine 1, the outputs of the above-mentioned sensors are input to the electronic control unit 30.
FIG. 2 shows an outline of the hardware configuration of the electronic control unit 30. Next, the internal configuration of the electronic control unit 30 will be described with reference to FIG.

As shown in FIG. 2, the electronic control unit 30
Is mainly composed of a microcomputer 31 including a CPU 31a, a ROM 31b, a RAM 31c, a backup RAM 31d and the like.

At the input / output port of the microcomputer 31, the idle switch 8b, the rotation speed sensor 15,
The cylinder discrimination sensor 16, the igniter 12, the heater energization control circuit 32, the drive circuit 35, etc. are connected.

Here, the heater energization control circuit 32 uses the battery 40 as a power source and the heater 11b of the oxygen sensor 11.
It is a circuit that controls the electric power supplied to the. Oxygen sensor 11
In this way, the detection element 11a is heated and activated by the heater 11b whose energization is controlled in this way.

Further, the drive circuit 35 includes the fuel injection valve 1 described above.
This is a circuit for driving 0. The input / output port of the microcomputer 31 has an A / D conversion circuit 34.
The sensors for outputting analog signals such as the intake pressure sensor 7, the intake temperature sensor 9, the water temperature sensor 17, and the opening degree sensor 8a are connected via the. Further, the output of the heater energization control circuit 32, the voltage across the current detection resistor 33, and the output of the detection element 11a of the oxygen sensor 11 are also input to the A / D conversion circuit 34.

The electronic control unit 30 executes various controls for fuel injection and ignition of the internal combustion engine 1 based on the outputs of the respective sensors thus captured by the microcomputer 31.

FIGS. 3 and 4 functionally show a control structure mainly relating to the air-fuel ratio feedback control and the air-fuel ratio learning control of the electronic control unit 30.
Referring to FIG. 3 and FIG. 4 together, the electronic control unit 30
The basic functions for executing these controls are described.

First, in the electronic control unit 30 shown in FIG. 3, the basic injection amount calculation unit 301 is based on the output NE of the rotation speed sensor 15 and the output PM of the intake pressure sensor 7 among the sensor outputs taken in as described above. Is a part for calculating the basic fuel injection amount TP to the engine 1. For this calculation, for example, the same basic fuel injection amount TP that is adapted to correspond to each operation range determined according to each value of the rotational speed NE and the intake pressure PM.
It is possible to use a map in which the value of is previously registered in the memory.

Further, in the electronic control unit 30, the air-fuel ratio correction coefficient calculation unit 302 sets the output R (rich) / of the oxygen sensor 11 under the condition that the air-fuel ratio feedback condition is satisfied as the condition A. This is a part for reading L (lean) and calculating an air-fuel ratio correction coefficient FAF according to the output.

Here, the air-fuel ratio feedback condition as the condition A is, for example, that (A1) the cooling water temperature (output of the water temperature sensor 17) is 80 ° C. or higher. (A2) Fuel is not being cut. (A3) The fuel acceleration amount is not being increased. Are all satisfied.

The FAF memory 303 is a memory for temporarily storing the air-fuel ratio correction coefficient FAF calculated in this way. These temporarily stored air-fuel ratio correction factors FAF
Is used as a correction coefficient of the air-fuel ratio feedback control described later, and is also used as a learning element of the learning control unit 304 described below.

The learning control section 304 is a section for learning the deviation of the air-fuel ratio for each operating region in accordance with the operating state of the engine 1, provided that the learning condition is satisfied as the condition B. . The specific structure of the learning control unit 304 will be described in more detail with reference to FIG.

That is, the learning control unit 304 shown in FIG.
In the above, the learning condition determination unit 411 is a part that determines whether or not the condition B, which is a learning condition, is satisfied, and the N counter 412 is a counter that counts the number N of times of taking in the air-fuel ratio correction coefficient FAF. This N counter 412
In the learning execution (learning value update) time, each time the learning condition determination unit 411 determines that the learning condition is satisfied, the determination unit 411 increments (+1) the count value N thereof.

Further, the air-fuel ratio correction coefficient average value calculation unit 413
Is the average value FAFAV of the air-fuel ratio correction coefficient FAF,
For example, it is a part calculated as FAFAV = (FAFi-1 + FAFi) / 2 (1). Here, FAFi-1 indicates the air-fuel ratio correction coefficient FAF previously fetched, and FAFii
Indicates the air-fuel ratio correction coefficient FAF taken in this time. In this embodiment, only the skip-corrected air-fuel ratio correction coefficient FAF, which will be described later, is used for the learning here.

Further, the deviation δ calculation unit 414 calculates the above value on the condition that the count value (the number of times the air-fuel ratio correction coefficient FAF is taken in) N of the N counter 412 reaches a predetermined maximum value (learning value update threshold value) Nmax. This is a part for obtaining the deviation δ from the theoretical value (for example, “1.0”) of the averaged value FAFAV. That is, if this theoretical value is "1.0",
The deviation δ is calculated as δ = 1.0−FAFAV (2) In the deviation δ calculation unit 414, the learning condition is satisfied immediately after the learning condition is determined through the learning condition determination unit 411, that is, even if the current learning condition is not satisfied, the learning condition is satisfied at the previous learning execution time. If the determination is made, the calculation of the equation (2) is executed based on the air-fuel ratio correction coefficient average value FAFAV immediately before that.

The learning value update amount calculation unit 415 is a part that calculates the learning value update amount DFAF according to the calculated deviation δ. Here, for example, δ ≦ −0.15 → DFAF = −0.05−0.08 ≧ δ> −0.15 → DFAF = −0.01−0.03 ≧ δ> depending on the value of the deviation δ. -0.08 → DFAF = -0.002 +0.03>δ> -0.03 → DFAF = 0 + 0.03 ≦ δ <+0.08 → DFAF = + 0.002 + 0.08 ≦ δ <+0.15 → DFAF = + 0.01 δ ≧ + 0.15 → DFAF = + 0.05 (3) The learning value update amount DFAF is calculated and set.

Further, when it is determined that the learning value update amount correction unit 416 has just deviated from the above-mentioned learning condition, the learning set by the above calculation according to the count value N of the N counter 412 at that time. This is a part for correcting the value update amount DFAF. Here, for example, according to the count value N, N ≦ N1 → DFAF ′ = DFAF × α1 N1 <N ≦ N2 → DFAF ′ = DFAF × α2 N> N2 → DFAF ′ = DFAF × α3 (4) , The learning value update amount correction value DFAF ′ is obtained. Here, N1, N2, and Nmax have a relationship of N1 <N2 <Nmax, and the correction coefficients α1, α2, and α3 have a relationship of α1 <α2 <α3 <1.

When it is determined that the learning condition is satisfied, the learning value calculation unit 417 determines the learning value update amount DFA.
FRAN = FRAN + DFAF (5) based on F, and when it is determined that the learning condition has just been deviated, the learning value update amount correction value DFA is set.
Based on F ′, it is a part that updates and calculates the learning value FRAN in the form of FRAN = FRAN + DFAF ′ (6). Further, in this learning value calculation unit 417, after updating the learning value FRAN in this way, the N counter 41
The count value N of 2 is reset to "0". By resetting the N counter 412 in this way, a new learning element (air-fuel ratio correction coefficient FAF)
Alternatively, new learning is started based on the average value FAFAV).

As described above, according to the learning control section 304, learning with good followability based on the deviation δ from the theoretical value of the air-fuel ratio correction coefficient average value FAFAV is executed, and
Even if the learning condition is deviated from before the air-fuel ratio correction coefficient FAF is fetched a predetermined number of times and immediately after that, the correction coefficient FF
Learning is executed according to the number of times AF is captured.

The learning condition as the condition B is as follows.
For example, (B1) the air-fuel ratio feedback control is being performed. (B2) The cooling water temperature (output of the water temperature sensor 17) is 80 ° C. or higher. (B3) The amount of increase after starting is “0”. (B4) The warm-up increase amount is “0”. Are all satisfied.

The calculated or updated learning value FRAN is stored in the FRAN memory 305 which is a learning value memory.
Stored in. The FRAN memory 305 is provided in the backup RAM 31d in which the stored value is backed up by the battery 40, as a memory having a storage area corresponding to each operation area.

In the electronic control unit 30 shown in FIG. 3, the fuel injection amount setting unit 306 is stored in the FAF memory 303 together with the basic fuel injection amount TP calculated by the basic injection amount calculation unit 301. Air-fuel ratio correction factor F
A portion for executing the calculation of the following equation based on AF and the learning value FRAN stored in the FRAN memory 305 corresponding to the current operating state to set the final fuel injection amount TAU. Is. TAU = TP * FAF * FRAN * FALL (7) Here, FALL does not depend on the air-fuel ratio correction coefficient FAF and the learning value FRAN, for example, the intake temperature correction coefficient, the transient correction coefficient, the battery voltage correction coefficient, and the like. It is another correction coefficient.

In the drive circuit 35, the fuel injection valve 10 is driven for the time corresponding to the fuel injection amount TAU based on the fuel injection amount TAU calculated and set in this way. 5 to 8 show an electronic control unit 30 having such a control structure.
The above-mentioned air-fuel ratio feedback control and learning control (learning value update) are shown as a series of processing procedures. The series of processing will be described in more detail below with reference to FIGS. To do.

(Air-fuel ratio feedback control) First, referring to FIG.
The control procedure of the air-fuel ratio feedback control will be described based on FIG. This control is performed by the microcomputer 31 (CPU 31a) in 16 ms (millisecond).
It is executed as a timer interrupt for each.

In the control routine shown in FIG. 5, the electronic control unit 30 firstly proceeds to step S101.
Detection signals from the various sensors described above (for example, the rotational speed NE,
Intake pressure PM, cooling water temperature THW, etc.) are read. And
In the following step S102, the basic injection amount calculation unit 30
1, the basic fuel injection amount TP corresponding to the rotational speed NE, the intake pressure PM, etc. is calculated.

Next, the electronic control unit 30 determines in step S
At 103, it is detected whether the air-fuel ratio feedback condition is satisfied. The air-fuel ratio feedback condition means that the condition (A1) to (A3) shown as condition A, that is, (A1) the cooling water temperature (the output of the water temperature sensor 17) is 80 ° C. or higher. (A2) Fuel is not being cut. (A3) The fuel acceleration amount is not being increased. Is a logical product condition of.

When this air-fuel ratio feedback condition is not satisfied, the air-fuel ratio correction coefficient FAF is set to "1.0" in step S104, and the process proceeds to step S130. Incidentally, in this case, the air-fuel ratio is not corrected.

On the other hand, when this air-fuel ratio feedback condition is satisfied, the electronic control unit 30 determines in step S
At 110, the output of the oxygen sensor 11 is read, and at step S111, this is R (rich) / L (lean).
Which of the two is shown is determined. In this determination, for example, the voltage corresponding to the theoretical air-fuel ratio (λ = 1) is referred to as the determination reference voltage.

When it is determined in step S111 that the output of the oxygen sensor 11 is L (lean), the electronic control unit 30 further determines in step S112.
It is determined whether this is the first L (lean). That is, it is determined whether or not the determination result of this time is reversed from R (rich) to L (lean) by comparing with the determination result of the previous time. And this time, from R (rich) to L (lean)
When it is determined that the value has been reversed, the FAF that is FAF ← FAF + A (A: skip amount) is calculated as a new air-fuel ratio correction coefficient FAF in step S113. On the other hand, if it is already L (lean) last time and it is judged that it is not newly reversed from R (rich) this time, in step S114, FAF ← FAF + a (a: integral amount, A > A) is calculated as a new air-fuel ratio correction coefficient FAF.

When it is determined in step S111 that the output of the oxygen sensor 11 is R (rich), the electronic control unit 30 determines in step S115.
It is determined whether this is the first R (rich). And
If it is newly determined that L (lean) is reversed to R (rich) this time, in step S116, FAF ← FAF-B (B: skip amount) is changed to a new air-fuel ratio correction coefficient FAF. Calculate as On the other hand, if it is already R (rich) in the previous time and it is determined that it is not newly inverted from L (lean) this time, in step S108, FAF ← FAF-b (b: integration amount , B> b) is calculated as a new air-fuel ratio correction coefficient FAF.

By performing such an arithmetic processing for the air-fuel ratio correction coefficient FAF, R (rich) and L
If there is an inversion with (lean), the same correction coefficient F
A large correction (skip) for AF is performed, and R
When the determination of (rich) or L (lean) is maintained, a comparatively gentle correction (integration) is performed with the same correction coefficient FA.
It will be done for F.

The air-fuel ratio correction coefficient FAF thus calculated is sequentially stored in the FAF memory 303,
Further, the skip-corrected air-fuel ratio correction coefficient FAF
Only this will be used for the learning control (learning value update) in the next step S120.

In step S130, the air-fuel ratio correction coefficient FAF calculated in this way, and in step S120.
The final fuel injection amount TAU is set based on the equation (7) based on the learning value FRAN updated in. As described above, the fuel injection valve 10 is driven through the drive circuit 35 only for the time corresponding to the fuel injection amount TAU thus set.

(Learning Control (Learning Value Update)) Next, the control procedure of the learning control (learning value update) mainly executed by the learning control unit 304 of the electronic control unit 30 as the processing of step S120 will be described. A description will be given based on FIGS. 6 to 8.

That is, in the learning value updating routine shown in FIG. 6, the electronic control unit 30 firstly executes step S201.
At, it is detected whether or not the learning condition is satisfied. This learning condition is defined as the condition B (B1) to (B).
The condition 4), that is, (B1) the air-fuel ratio feedback control is being performed. (B2) The cooling water temperature (output of the water temperature sensor 17) is 80 ° C. or higher. (B3) The amount of increase after starting is “0”. (B4) The warm-up increase amount is “0”. Is a logical product condition of.

If any one of these conditions is not satisfied, the electronic control unit 30 determines that the learning condition is not satisfied, and in step S202, determines that the learning condition is not satisfied last time. If so, step S2
At 03, the N counter 412 is reset and the routine is temporarily exited.

On the other hand, when it is determined in step S201 that the learning condition is satisfied, the electronic control unit 30
After incrementing (+1) the count value N of the N counter 412 in step S210,
At 1, the average value FAFAV of the air-fuel ratio correction coefficient FAF is calculated based on the above equation (1).

In this case, in step S212,
The incremented count value N and the predetermined maximum value N
Max is compared, and the routine is once exited until the count value N reaches Nmax.

When it is determined in step S212 that the count value N has reached the value Nmax, the electronic control unit 30 further updates the corresponding learning value FRAN in the following procedure. To do.

That is, when updating the learning value, the electronic control unit 30 firstly proceeds to step S213 based on the above equation (2) to calculate the theoretical value (“1.0”) of the average value FAFAV.
Is calculated, and in the next step S214, the learning value update amount D corresponding to this deviation δ is calculated based on the above equation (3).
Calculate and set FAF. The calculation procedure of the learning value update amount DFAF based on the equation (3) is shown in FIG.
This is as indicated by S308.

The electronic control unit 30 having calculated the learning value update amount DFAF in this way next updates the learning value FRAN based on the above equation (5) in step S215, and further, in step S216, the N counter 412. After resetting the count value N of (N = 0), the routine is once exited.

On the other hand, when it is determined in step S202 that the learning condition is not satisfied this time but is satisfied last time, the electronic control unit 30 responds in the following manner. The learning value FRAN is updated.

That is, in this case, the electronic control unit 30 first executes, in step S221, the above equation (2) based on the air-fuel ratio correction coefficient average value FAFAV previously obtained, and the theoretical value (" 1.0 ").
Then, in step S222, the learning value update amount DFAF corresponding to the obtained deviation δ is calculated based on the same equation (3) (FIG. 7) as above, and then in step S223, the calculated update amount DFAF is calculated as described above. Correction is performed according to the count value N of the N counter 412 at that time. This correction mode is as shown in the above equation (4), and the correction procedure is illustrated in FIG.

That is, as shown as a learning value update amount DFAF correction routine in FIG.
In the (learned value update amount correction unit 416), the N counter 412
After reading the count value N at that time (step S40
1) If the count value N is equal to or less than the value “N1” (step S402), the calculation DFAF ′ = DFAF × α1 is executed (step S403), and the count value N is the value “N”.
If it is larger than "1" and equal to or smaller than the value "N2" (step S404), the operation DFAF '= DFAF × α2 is executed (step S405), and if the same count value N is larger than the value "N2" ( In step S404, the calculation DFAF ′ = DFAF × α3 is executed (step S406) to calculate the correction value DFAF ′ for the calculated update amount DFAF. At this time, the values “N1” and “N2” have a relationship of N1 <N2 <Nmax, and the correction coefficients α1, α2, and α3 have a relationship of α1 <α2 <α3 <1. That is, when the number of times (N) of taking in the air-fuel ratio correction coefficient FAF is small (for example, when N ≦ N1), the average value FAF
Since the reliability of AV is low, its correction coefficient (for example, α1)
Also, a smaller value is used.

The electronic control unit 30 having corrected the learned value update amount DFAF in this way next updates the learned value FRAN based on the above equation (6) in step S224 (FIG. 6). Then, in step S225, the N counter 41
After resetting the count value N of 2 (N = 0), the routine is temporarily exited.

As described above, according to the air-fuel ratio control apparatus of the first embodiment, the learning control section 304 provides good followability based on the deviation δ from the theoretical value of the air-fuel ratio correction coefficient average value FAFAV. Learning is carried out.

Further, even if the learning condition is deviated from before the air-fuel ratio correction coefficient FAF is fetched a predetermined number of times, immediately after that, based on the update amount DFAF 'corrected in accordance with the number of fetches of the correction coefficient FAF. , Learning with good followability is maintained.

Therefore, the chances of updating each learning value FRAN according to the operating state of the internal combustion engine 1 are surely increased, and more precise air-fuel ratio control that responds immediately to each state of the engine 1 is realized. Will be done.

In the apparatus of the first embodiment, the learning value update amount DFAF is corrected in three steps in the correction processing of step S223 of the learning value update routine, but this is only an example. It is arbitrary how much resolution and how much correction is performed.

Further, in the apparatus of the first embodiment, the learning value update amount DFA obtained based on the deviation δ in the correction processing of step S223 of the learning value update routine is also performed.
Although F is corrected according to the number of times N of taking in the air-fuel ratio correction coefficient FAF, the learning value update amount according to the number of times of taking-in N can be directly obtained without referring to the deviation δ.

Similarly, in the learning value updating routine, in the apparatus of the first embodiment, when the learning condition is satisfied, the number N of times of taking in the air-fuel ratio correction coefficient FAF is a predetermined value Nm.
Although the control structure that updates the learning value when the ax value is reached is adopted, the control structure may always update the learning value when the learning condition is satisfied. Of course, even in this case, immediately after the learning condition is deviated, the same air-fuel ratio correction coefficient F
It is possible to adopt a control structure in which the learning value update amount is set according to the number N of times of taking in AF.

The control structure in which the learning value update amount is set according to the number N of times the air-fuel ratio correction coefficient FAF is taken in is applied not only immediately after the learning condition is deviated but also when the learning condition is satisfied. be able to.

Further, in the apparatus of the first embodiment, the N counter 412 is provided and the intake of the air-fuel ratio correction coefficient FAF as a learning element is managed by the number of times N. It is also possible to provide an appropriate timing means and manage the intake of the air-fuel ratio correction coefficient FAF by the time.

(Second Embodiment) Next, a second embodiment of the air-fuel ratio control device according to the present invention will be described. The second
The device of the embodiment of the present invention is a device that also uses learning control for air-fuel ratio control of an internal combustion engine mounted on an automobile, etc., and the air-fuel ratio correction coefficient is adjusted according to the deviation amount between the average value and the theoretical value. It is configured as an apparatus having a control structure in which the number of times the fuel ratio correction coefficient is taken in is changed.

However, even in the device of the second embodiment, the structure of the internal combustion engine to which the invention is applied and its peripheral devices, and the structure shown in FIG. 2 or FIG. The first embodiment of the present invention is similar to the apparatus of the first embodiment of the present invention, and redundant description of these configurations will be omitted here.

The apparatus of the second embodiment is, for example, the learning control unit 304 shown in FIG. 3 and has a control structure shown as the learning control unit 304 'in FIG. Has become.

That is, in the learning control unit 304 'shown in FIG. 9, the learning condition judging unit 421 is a unit for judging whether or not the condition B, which is a learning condition, is satisfied.
The N counter 422 is a counter that counts the number N of times of taking in the air-fuel ratio correction coefficient FAF. This N counter 422 also increments its count value N (+1) through the determining unit 421 each time the learning condition determining unit 421 determines that the learning condition is satisfied during the learning execution (learning value update) timing. To be done.

Further, the air-fuel ratio correction coefficient average value calculation unit 423
Is the average value FAFAV, provided that the number N of times of taking in the air-fuel ratio correction coefficient FAF has reached the minimum value Nmin (for example, "2" or "4") at which the average value of the correction coefficient FAF can be calculated. Is a part that calculates based on the above equation (1), and the deviation δ calculation unit 424 calculates the deviation δ from the theoretical value of the calculated average value FAFAV based on, for example, the above expression (2). Is.

In the learning control unit 304 ', N
The max setting unit 425 uses the Nman determination flag (memory).
On condition that the flag is not set in 426,
This is a part for setting the maximum value (learning value update threshold) Nmax of the number N of times of capturing according to the calculated deviation δ.

The maximum value Nmax is set as follows: (a) When the deviation δ is large, the air-fuel ratio correction coefficient FAF
Is estimated to be largely deviated from the theoretical value, so a small value is set as the value of Nmax in order to correct this at an early stage. (B) When the deviation δ is small, the air-fuel ratio correction coefficient FAF
Since it is estimated that the difference between the theoretical value and the theoretical value is small, a large value is set as the value of Nmax in order to perform more accurate correction. Shall be performed based on an algorithm such as.

Further, in the Nmax setting unit 425, when the setting of the maximum value Nmax is completed in this way, it is set to Nmax.
The Nman determination flag (memory) 426 is stored in the memory 427 and set in the Nman determination flag (memory) 426.

Then, in the learning control unit 304 ', the learning value calculation unit 428 determines the N counter 4 based on the maximum value Nmax stored in the Nmax memory 427.
The learning value FRAN is updated in a mode such as FRAN = FRAN + C (8) on condition that the count value N of 22 reaches the maximum value Nmax. In the equation (8), for example, a value of about “± 0.01” is used as the update amount “C”.

Further, in the learning value calculation unit 428, after updating the learning value FRAN in this way, the N counter 42
The count value N of 2 is reset to "0" and
The same flag in the man determination flag (memory) 426 is cleared. Thus, the N counter 422 is reset and N
By clearing the max determination flag 426, new learning based on a new learning element (air-fuel ratio correction coefficient FAF or its average value FAFAV) will be started when subsequent learning conditions are satisfied.

As described above, according to the learning control unit 304 ', the maximum value (learning value update threshold) Nmax for the number of times N the air-fuel ratio correction coefficient FAF is taken in is calculated from the theoretical value of the air-fuel ratio correction coefficient average value FAFAV. It will be suitably adjusted and changed based on the deviation δ of.

The learning condition as the condition B is, for example, (B1) that the air-fuel ratio feedback control is in progress. (B2) The cooling water temperature (output of the water temperature sensor 17) is 80 ° C. or higher. (B3) The amount of increase after starting is “0”. (B4) The warm-up increase amount is “0”. The setting as the logical product of is similar to the case of the first embodiment.

The updated learning value FRAN is
It is stored in the FRAN memory 305 (FIG. 3) which is a learning value memory, and one of the stored learning values FRAN corresponding to the current operating state and the FAF memory 3 are stored.
03 (FIG. 3), the air-fuel ratio correction coefficient FAF,
And the final fuel injection amount TAU is set based on the calculation of the equation (7) based on the basic fuel injection amount TP calculated based on the rotational speed NE and the intake pressure PM, and so on.
This is the same as the case of the embodiment.

FIG. 10 shows a series of processing procedures for learning control (learning value update) of the apparatus of the second embodiment. Hereinafter, with reference to FIG. The processing according to the above will be described in more detail. Even in the device of the second embodiment, the air-fuel ratio feedback control itself is the same as that of the device of the first embodiment, and the description of the control procedure will be omitted.

(Learning Control (Learning Value Update)) In the learning value updating routine shown in FIG. 10, the electronic control unit 30 first detects in step S501 whether or not the learning condition (condition B) is satisfied. To do. Then, the above condition (B
If any one of 1) to (B4) is not satisfied, the electronic control unit 30 determines that the learning condition is not satisfied,
After resetting the N counter 422 in step S502, the routine is temporarily exited.

On the other hand, if it is determined in step S501 that the learning condition is satisfied, the electronic control unit 30
After incrementing (+1) the count value N of the N counter 422 in step S510, the step S51
At 1, the incremented count value N is the above-mentioned minimum value N.
It is determined whether or not min has been reached. If the count value N has not reached the minimum value Nmin yet, the routine is temporarily exited.

If the same count value N has reached the minimum value Nmin, the average value FAFAV of the air-fuel ratio correction coefficient FAF is calculated based on the above equation (1) in step S512, and then in step S513, Based on the equation (2), the deviation δ from the theoretical value of the average value FAFAV is calculated.

The electronic control unit 3 which calculates the deviation δ in this way
Next, in step S514, it is determined whether or not the Nmax determination flag 426 is set. If not, in step S515, (a)
Based on the algorithms (b) and (b), the maximum value (learning value update threshold value) Nmax for the number of times N the air-fuel ratio correction coefficient FAF is taken in is set according to the deviation δ. In addition, in the next step S516, the Nmax determination flag 426 is set.

When the Nmax decision flag 426 has already been set, or in this way, the number of fetches N
If the maximum value Nmax is set and the same Nmax determination flag 426 is set, the electronic control unit 30 further determines in step S517 whether or not the count value N has reached the set maximum value Nmax. to decide. When it is determined that the count value N has not reached the maximum value Nmax yet, the routine is temporarily exited.

On the other hand, if it is determined in step S517 that the count value N has reached the maximum value Nmax,
In step S518, electronic control unit 30 updates the corresponding learning value FRAN based on the above equation (8). Then, in step S519, the N counter 422
The count value N of N is reset (N = 0), and step S520
After clearing the Nmax determination flag 426, the routine is temporarily exited.

As described above, according to the air-fuel ratio control system of the second embodiment, the learning control section 304 '
The maximum value (learning value update threshold) Nmax for the number of times N the air-fuel ratio correction coefficient FAF is taken in is the same as the correction coefficient average value F.
It can be set efficiently based on the deviation δ from the theoretical value of AFAV.

Assuming that the upper limit of this value Nmax is set to the same degree as the number of times of capture adopted as the threshold in the conventional learning control structure, the apparatus of the second embodiment also The chance that each learning value FRAN corresponding to the operating state of the internal combustion engine 1 is updated will surely increase. That is, more accurate air-fuel ratio control that responds to the respective conditions of the engine 1 is realized.

In the apparatus of the second embodiment as well, the N counter 422 is provided and the intake of the air-fuel ratio correction coefficient FAF as a learning element is managed by the number of times N, but in addition, for example, It is also possible to provide an appropriate timing means and manage the intake of the air-fuel ratio correction coefficient FAF by the time.

Further, in the second embodiment, as shown in the equation (8), the learning value FRAN is calculated (updated) by
The fixed update amount “C” is used, but the selection is arbitrary. That is, also for this update amount, it is possible to use the variable update amount DFAF calculated according to the deviation δ, for example, in the manner shown in the equation (3), as in the first embodiment. .

In addition, a learning control structure in which the first and second embodiments are combined is adopted, for example, the learning value update amount is further corrected according to the number of times the air-fuel ratio correction coefficient FAF is taken in or the time taken. Of course you can do it.

In each of the first and second embodiments, the skip-corrected air-fuel ratio correction coefficient FA is used.
Only F was decided to be used for such learning. But,
If the average value FAFAV is always calculated so as to be favorably compared with the theoretical value, all the correction coefficients FAF including the air-fuel ratio correction coefficient FAF that is integral-corrected may be incorporated as learning elements. it can. In this case, it is expected that the opportunity for updating the learning value will be further increased.

Further, in each of the learning control structures described above, the method of calculating the air-fuel ratio correction coefficient FAF itself is arbitrary, and various well-known methods are not necessarily limited to the method illustrated in FIG. You can

[0123]

As described above, according to the present invention, the opportunity to update the learning value is properly and surely increased, and the more accurate air-fuel ratio that immediately responds to the engine state at each time. Control comes to be realized.

[Brief description of drawings]

FIG. 1 is a block diagram showing a first embodiment of an air-fuel ratio control device according to the present invention.

FIG. 2 is a block diagram showing the configuration of an electronic control unit of the embodiment.

FIG. 3 is a block diagram functionally showing an air-fuel ratio control structure of the electronic control unit.

FIG. 4 is a block diagram functionally showing a learning control structure of the electronic control unit.

FIG. 5 is a flowchart showing an air-fuel ratio feedback control procedure of the embodiment.

FIG. 6 is a flowchart showing a learning value updating procedure of the embodiment.

FIG. 7 is a flowchart showing a learning value update amount calculation procedure of the embodiment.

FIG. 8 is a flowchart showing a learning value update amount correction procedure of the embodiment.

FIG. 9 is a block diagram functionally showing a learning control structure of a second embodiment of the air-fuel ratio control device of the present invention.

FIG. 10 is a flowchart showing a learning value updating procedure of the second embodiment.

[Explanation of symbols]

1 ... Internal combustion engine, 2 ... Intake system, 3 ... Combustion chamber, 4 ... Exhaust system,
5 ... Throttle valve, 6 ... Surge tank, 7 ... Intake pressure sensor, 8 ... Throttle position sensor, 9 ... Intake temperature sensor, 10 ... Fuel injection valve, 11 ... Oxygen sensor (air-fuel ratio sensor), 12 ... Igniter, 13 ... Distributor, 14 ... Spark plug, 15 ... Rotation speed sensor, 16 ... Cylinder discrimination sensor, 17 ... Water temperature sensor, 30 ... Electronic control device, 31 ... Microcomputer, 31a ... CPU, 3
1b ... ROM, 31c ... RAM, 31d ... Backup RAM, 32 ... Heater energization control circuit, 33 ... Current detection resistor, 34 ... A / D conversion circuit, 35 ... Drive circuit, 40
... Battery, 301 ... Basic injection amount calculation unit, 302 ... Air-fuel ratio correction coefficient calculation unit, 303 ... FAF memory, 304 ... Learning control unit, 305 ... FRAN memory, 306 ... Fuel injection amount setting unit, 411,421 ... Learning condition Determination unit, 412,
422 ... N counter, 413, 423 ... Air-fuel ratio correction coefficient average value calculation unit, 414, 424 ... Deviation δ calculation unit, 415
Learning value update amount calculation unit, 416 Learning value update amount correction unit,
417, 428 ... Learning value calculation unit, 425 ... Nmax setting unit, 426 ... Nmax determination flag, 427 ... Nmax memory.

Claims (6)

[Claims] 1. A fuel injection valve for injecting fuel pressure-fed from a fuel tank into a combustion chamber of an internal combustion engine, and an air-fuel ratio provided in an exhaust system of the engine for detecting an air-fuel ratio of the engine from exhaust gas thereof. Detection means, air-fuel ratio correction coefficient calculation means for calculating an air-fuel ratio correction coefficient according to the detected air-fuel ratio, and this calculated air-fuel ratio correction so that the air-fuel ratio of the engine converges within a predetermined range. Feedback control means for feedback controlling the operation amount of the fuel injection valve based on a coefficient, and based on the calculated air-fuel ratio correction coefficient, while changing the update amount according to the number of times of intake or the time of intake, the engine concerned Learning control means for learning the air-fuel ratio correction amount according to the operating state of the vehicle, and correcting the operation amount of the fuel injection valve that is feedback-controlled according to the learned air-fuel ratio correction amount. Air-fuel ratio control system for an internal combustion engine, characterized in that it comprises a correction unit that, the. 2. The learning control means calculates an average value of the fetched air-fuel ratio correction coefficients, and a deviation amount calculation means for calculating a deviation amount between the calculated average value and a theoretical value. And an update amount calculation means for calculating the update amount of the air-fuel ratio correction amount according to the operating state of the engine, which is a learning value according to the calculated deviation amount, and the number of times the intake of the air-fuel ratio correction coefficient or the intake time is taken. 2. The internal combustion engine according to claim 1, further comprising: an update amount correction unit that corrects the calculated update amount in accordance with the above; and a learning value calculation unit that updates a learning value based on the corrected update amount. Air-fuel ratio control system for engines. 3. The learning control means, a learning condition judging means for judging whether or not a learning condition is satisfied, and an average value of the fetched air-fuel ratio correction coefficients on condition that the learning condition is satisfied. An average value calculating means for calculating, and on condition that the number of times of taking in the air-fuel ratio correction coefficient or the taking time reaches a predetermined number of times or a predetermined time
A first deviation amount calculating means for calculating a first deviation amount between the calculated average value and the theoretical value, and an operating state of the engine which is a learning value according to the calculated first deviation amount. First update amount calculation means for calculating a first update amount of the corresponding air-fuel ratio correction amount, and first learning value calculation means for updating a learning value based on the calculated first update amount, A second deviation amount calculation means for calculating a second deviation amount between the average value and the theoretical value calculated immediately before the air-fuel ratio correction coefficient immediately after the learning condition is not satisfied; Second update amount calculation means for calculating a second update amount of the air-fuel ratio correction amount according to the operating state of the engine, which is a learned value according to the deviation amount, and the number of times of taking in the air-fuel ratio correction coefficient at that time Alternatively, an update for correcting the calculated second update amount according to the acquisition time A correction unit, an air-fuel ratio control system for the corrected second second learned value calculating means and the internal combustion engine according to claim 1 constituted comprising the updating of the learning value based on the update amount. 4. A fuel injection valve for injecting fuel pressure-fed from a fuel tank into a combustion chamber of an internal combustion engine, and an air-fuel ratio which is provided in an exhaust system of the engine and detects an air-fuel ratio of the engine from exhaust gas thereof. Detection means, air-fuel ratio correction coefficient calculation means for calculating an air-fuel ratio correction coefficient according to the detected air-fuel ratio, and this calculated air-fuel ratio correction so that the air-fuel ratio of the engine converges within a predetermined range. Feedback control means for feedback-controlling the operation amount of the fuel injection valve based on a coefficient, and changing the number of times of intake or the time of intake in accordance with the calculated air-fuel ratio correction coefficient, while changing the empty state according to the operating state of the engine. Learning control means for learning the fuel ratio correction amount; correction means for correcting the operation amount of the fuel injection valve that is feedback-controlled according to the learned air-fuel ratio correction amount; Air-fuel ratio control system for an internal combustion engine, characterized in that it comprises. 5. The learning control means calculates an average value of the fetched air-fuel ratio correction coefficients, and a deviation amount calculation means for calculating a deviation amount between the calculated average value and a theoretical value. And a threshold setting means for setting a threshold value for the number of times of taking in the air-fuel ratio correction coefficient or the taking-in time according to the calculated deviation amount, and the number of times of taking-in the air-fuel ratio correction coefficient or the taking-in time. The learning value calculation means for updating the air-fuel ratio correction amount corresponding to the operating state of the engine, which is a learning value, on the basis of a predetermined update amount on condition that the threshold value is reached, Air-fuel ratio control device for internal combustion engine. 6. The air-fuel ratio control apparatus for an internal combustion engine according to claim 5, wherein an update amount of the air-fuel ratio correction amount corresponding to the operating state of the engine, which is the learning value, is calculated according to the calculated deviation amount. The air-fuel ratio control apparatus for an internal combustion engine, further comprising: an update amount calculating means for updating the learned value as the predetermined update amount based on the calculated update amount.