JPS61190142A - Learning control device of internal-combustion engine - Google Patents

Abstract

PURPOSE:To enable highly accurate and high speed learning by providing a deviation means value arithmetic means calculating means value from upper and lower peak values of deviation and a new learning correcting amount arithmetic means calculating new learning correcting amount based on the mean value. CONSTITUTION:A learning correcting amount correcting means H learns deviation from basic value of feedback correcting amount during normal condition detecting and corrects learning correcting amount and rewrites. A new learning correcting amount arithmetic means H3 adds means value of deviation obtained by a deviation means value arithmetic means H2 to the present learning correcting value by predetermined ratio to calculate new learning correcting amount each time when the mean value is obtained. Thus, highly accurate and high speed learning is made possible.

Description

DETAILED DESCRIPTION OF THE INVENTION (Industrial Field of Application) The present invention relates to a learning control device for a feedback control system for controlling the air-fuel ratio, idle speed, etc. of an internal combustion engine.

<Prior Art> Conventional learning control devices for internal combustion engines include, for example, an air-fuel ratio learning control device disclosed in Japanese Patent Laid-Open No. 59-203828, and an idler learning control device disclosed in Japanese Patent Laid-Open No. 59-211738. There is a learning control device for the rotation speed.

Here, a base air-fuel ratio learning control device will be described as an example in the case where the air-fuel ratio is feedback-controlled to the stoichiometric air-fuel ratio, which is a control target value, in an internal combustion engine having an electronically controlled fuel injection device.

A fuel injection valve used in an electronically controlled fuel injection system is opened by a drive pulse signal applied in synchronization with the rotation of an engine, and is supposed to inject fuel at a predetermined pressure during the period of the valve opening. Therefore, the fuel injection amount is controlled by the pulse of the drive pulse signal, and if Ti is the control signal corresponding to the fuel injection amount during this pulse, Ti is determined by the following equation in order to obtain the stoichiometric air-fuel ratio.

Ti=Tp-COEF・α+Ts However, 'rp is called the basic fuel injection amount for convenience among the basic pulses corresponding to the basic fuel injection amount. 'rp=K・Q/N
where K is a constant, Q is the engine intake air flow rate, and N is the engine speed. C0EF is various correction coefficients such as water temperature correction.

α is a feedback correction coefficient for air-fuel ratio feedback control (λ control) to be described later. Ts is a voltage correction element, which is used to correct changes in the injection flow rate of the fuel injector due to fluctuations in battery voltage.

Regarding λ control, a 0□ sensor is installed in the exhaust system to detect the actual air-fuel ratio, determine whether the air-fuel ratio is richer or leaner than the stoichiometric air-fuel ratio based on the slice level, and adjust the fuel injection amount to achieve the stoichiometric air-fuel ratio. control, and for this reason,
Defining the feedback correction coefficient α mentioned above,
By changing this α, the stoichiometric air-fuel ratio is maintained.

Here, the value of the feedback correction coefficient α is the proportional integral (P
I) It is changed by control and stable control is achieved.

In other words, 0! Compare the sensor output voltage with the slice level voltage, and if it is higher or lower than the slice level, the air-fuel ratio will not suddenly become richer or leaner, and if the air-fuel ratio is rich (lean). First lower (raise) by P, then gradually lower (raise) 1 minute at a time.
and controls the air-fuel ratio to become leaner (richer).

However, under conditions where λ control is not performed, α is clamped and a desired air-fuel ratio is obtained by setting various correction coefficients COEF.

By the way, if the base air-fuel ratio under λ control conditions, that is, the air-fuel ratio when α = 1, could be set to the stoichiometric air-fuel ratio (λ = 1), feedback control would not be necessary. For example, variations in air flow meters, fuel injection valves, pressure regulators, control units), changes over time, and pulse width of fuel injection valves.
Feedback control is performed because the base air-fuel ratio deviates from λ=1 due to factors such as non-linearity of flow characteristics and changes in operating conditions and environment.

However, if the base air-fuel ratio deviates from λ=1, it takes time to stabilize the step in the base air-fuel ratio to λ=1 through feedback control when the operating range changes significantly. And for this we need the constant of proportionality and integration (P/
1 minute) increases, overshoot or undershoot occurs, resulting in poor controllability. In other words, if the base air-fuel ratio deviates from λ=1, the air-fuel ratio is controlled within a range that deviates considerably from the stoichiometric air-fuel ratio.

As a result, the three-way catalyst is operated at a point where its conversion efficiency is poor, and not only does the cost increase due to an increase in the amount of precious metal in the catalyst, but the conversion efficiency further deteriorates as the catalyst deteriorates, forcing the catalyst to be replaced. .

Therefore, by setting the base air-fuel ratio to λ = 1 through learning, it is possible to eliminate the deviation from λ = 1 caused by the step in the base air-fuel ratio during transients, and to reduce the P/I component, improving controllability. A learning control device for air-fuel ratio that aims to improve
Japanese Patent Application No. 58-76221 (Japanese Unexamined Patent Publication No. 58-76221)
9-203828) or patent application No. 1974-1974
It was filed as No. 9.

This is because if the base air-fuel ratio deviates from the stoichiometric air-fuel ratio during air-fuel ratio feedback control, the feedback correction coefficient α increases to fill the gap, so the engine operating state and α at this time are detected. , calculate the learning correction coefficient Kl based on the α and store it, and when the same engine operating condition returns, the base air-fuel ratio is corrected using the stored learning correction coefficient KJ so that it is responsive to the stoichiometric air-fuel ratio. do. The memory of the learning correction coefficient Kl here is R
The AM map is divided into grids according to appropriate parameters of the engine operating state such as engine speed and load, and this is done for each area within a predetermined range.

Specifically, a map of learning correction coefficients KJ corresponding to engine operating conditions such as engine speed and load is provided in the RAM, and when calculating the fuel injection amount Ti, the basic fuel injection 1tTp is learned as shown in the following formula. Correct using correction coefficient Kl.

T i =Tp-COEF-Kj2- ot+Ts Then, learning of Kl proceeds in the following steps.

i) Detect the region of the engine operating state at that time in a steady state, and detect the deviation Δα of α from the reference value α1 during that period.
(=α−α,) is detected. The reference value α is generally set to 1 as a value corresponding to λ=1.

ii) Searching for Kl that has been learned up to now corresponding to the region of the engine operating state.

1ii) Find the value of β+Δα/M from l(/ and Δα, and update the memory as the result (learning value) as a new Kl axis. M is a constant and M>1.

In addition, the idle speed learning control device includes an idle control valve in the auxiliary air passage that bypasses the throttle valve.
When controlling the idle speed by adjusting the opening degree of this idle control valve, the basic opening degree of the idle control valve corresponding to the target idle speed for each engine cooling water temperature is the target idle speed and the actual idle speed. When performing feedback correction while comparing the rotation speed, a learning correction amount map is created according to the cooling water temperature, which is a parameter of the engine operating state, and the learning correction amount is determined by learning the deviation from the reference value of the feedback correction amount. While making corrections, the basic opening degree is corrected using this learned correction amount to stabilize the control.

<Problems to be Solved by the Invention> By the way, when performing learning control, for example, learning control of air-fuel ratio, the deviation Δα of the feedback correction coefficient α from the reference value α is detected, and based on this the deviation Δα of the feedback correction coefficient α is determined. The coefficient KIl is updated, but if you capture the deviation from time to time as the deviation Δα and learn it each time, in the case of proportional-integral control, there is a change due to the P component, so if the P component is particularly large. Detection accuracy deteriorates in the region, and efficient learning cannot be performed.

Therefore, the deviation Δα is taken as an average value, that is, Δα is sampled for a certain predetermined time and divided by the number of samples,
A method was devised to calculate the average value.

However, if the sampling time is increased in order to increase accuracy, there is a problem in that it is disadvantageous to increasing the speed of learning control.

In view of these conventional problems, the present invention aims to improve the detection of the deviation of the feedback correction amount from the reference value when correcting the learning correction amount, thereby enabling highly accurate and high-speed learning control. With the goal.

(Means for Solving the Problems) In order to achieve the above object, the present invention, as shown in FIG. The amount correcting means H includes the following means H1 to H3.

(A) Basic control amount setting means for setting basic control amounts corresponding to control target values of internal combustion engine control objects such as air-fuel ratio and idle speed (B) Engines classified into multiple types according to parameters representing engine operating conditions A rewritable storage means (C) that stores learning correction amounts for correcting the basic control amount for each region of operating conditions; (C) retrieving learning correction amounts for the corresponding region from the storage means based on the actual engine operating state; Learning correction amount search means (D) Feedback for comparing the control target value and the actual value and increasing or decreasing the feedback correction amount by a predetermined amount to correct the basic control amount so that the actual value approaches the control target value. Correction amount setting means (F) The basic control amount set by the basic control amount setting means, the academic correction amount searched by the learning correction amount search means, and the feedback correction amount set by the feedback correction amount setting means. (F) A control means for calculating a control amount from the control amount (F) A control means for operating in accordance with the control amount to control control objects of the internal combustion engine such as the air-fuel ratio and idle speed (G) Actual engine operating state Steady state detection means (H) that detects that the is in a steady state. During steady state detection, the device learns the deviation of the feedback correction amount from the reference value during the steady state detection, and responds to the region of the engine operating state during that time in the direction of reducing it. A learning correction amount correcting means for correcting and rewriting a learning correction amount to be performed. Here, the learning correction amount correcting means H includes the following
including means of

(Old) Deviation peak value detection means that detects the peak value of the deviation from the reference value of the feedback correction amount at that time every time the direction of increase or decrease of the feedback correction amount is reversed during steady state detection (■2) This deviation peak value detection Deviation average value calculation means (H3) which calculates the average value from the upper and lower peak values of the deviation obtained by two detections before and after. New learning correction amount calculation means that calculates a new learning correction amount by adding a predetermined percentage of the average value of the deviation to the learning correction amount> The basic control amount setting means A sets control targets such as air-fuel ratio and idle rotation speed. The basic control amount corresponding to the value is set, for example, according to a predetermined calculation formula or by searching, and the learning correction amount retrieval means C retrieves the learning correction amount for the corresponding region from the storage means B based on the actual engine operating state. The feedback correction amount setting means compares the control target value and the actual value and sets the feedback correction amount by increasing or decreasing a predetermined amount based on, for example, proportional-integral control so that the actual value approaches the control target value. . Then, the control amount calculation means E calculates the control amount by correcting the basic control amount with the learning correction amount and further correcting it with the feedback correction amount, and the control means F operates according to this control amount, for example, the fuel The injection amount or auxiliary air amount is controlled to control the air-fuel ratio, idle speed, etc.

The steady state detection means G detects, for example, when the engine operating state is an arbitrary one.
A steady state is detected by continuously and stably existing in one region, and it is known that it is a state that can be learned.

When it is determined that the learning is possible, the learning correction amount is corrected/rewritten by the learning correction amount correction means H (H1 to H3).

That is, the deviation peak value detection means H1 detects the peak value of the deviation of the feedback correction amount from the reference value at that time every time the direction of increase/decrease of the feedback correction amount is reversed.

The deviation average value calculating means H2 calculates the average value from the upper and lower peak values of the deviation obtained by the two detections before and after.

Then, the new learning correction amount calculating means H3 calculates a new learning correction amount by adding the average value of the deviation to the current learning correction amount by a predetermined ratio every time the average value of the deviation is obtained. As a result, the learning correction amount is updated in a direction that reduces the deviation of the feedback correction amount from the reference value, and the learning correction amount data in the storage means B corresponding to the region of the engine operating state at this time is rewritten.

<Embodiment> An embodiment in which the learning control device of the present invention is applied to an air-fuel ratio feedback control system of an internal combustion engine having an electronically controlled fuel injection device will be described below.

In FIG. 2, the engine 1 includes an air cleaner 2°, an intake duct 3. Throttle chamber 4 and intake manifold 5
Air is inhaled through.

The intake duct 3 is provided with an air flow meter 6 as means for detecting the intake air flow rate Q, and outputs a voltage signal corresponding to the intake air flow rate Q signal.

The throttle chamber 4 includes a primary throttle valve 7 and a secondary throttle valve 8 that operate in conjunction with an accelerator pedal (not shown).
is provided to control the intake air flow rate Q. Further, an auxiliary air passage 9 is provided that bypasses these throttle valves 7.8, and an idle control valve 10 is interposed in this auxiliary air passage 9. Intake manifold 5
Alternatively, the intake boat of the engine 1 is provided with a fuel injection valve 11.

The fuel injection valve 11 is an electromagnetic fuel injection valve that opens when the solenoid is energized, and closes when the energization is stopped.The solenoid is energized by a drive pulse signal to open the valve, and pressure is supplied from a fuel pump (not shown). Fuel controlled to a predetermined pressure by a regulator is injected and supplied to the engine 1. Therefore, the fuel injection valve 11 is a control means for controlling the fuel injection amount by its operation and controlling the air-fuel ratio to the optimum air-fuel ratio (stoichiometric air-fuel ratio) which is a control target value.

From engine 1, exhaust manifold 12. Exhaust duct 13
．． Exhaust gas is discharged via the three-way catalyst 14 and the muffler 15.

The exhaust manifold 12 is provided with an 02 sensor 16. This 0□ sensor 16 is the oxygen concentration in the atmosphere (constant)
outputs a voltage signal according to the ratio of the oxygen concentration in the exhaust gas and
This is a known sensor whose electromotive force changes suddenly when the air-fuel mixture is combusted at the stoichiometric air-fuel ratio. Therefore, the 0□ sensor 16 is a means for detecting the air-fuel ratio (rich/lean) of the air-fuel mixture. The three-way catalyst 14 is a catalytic device that efficiently oxidizes or reduces Co, HC, and NOx in the exhaust gas near the stoichiometric air-fuel ratio of the air-fuel mixture and converts them into other harmless substances.

In addition, a crank angle sensor 17 is provided.

In the crank angle sensor 17, a signal disk plate 19 is provided on the crank pulley 18, and teeth provided on the outer circumference of the plate 19 output a reference signal every 120 degrees and a position signal every 1 degree, for example. Here, the engine speed N can be calculated by measuring the period of the reference signal.

Therefore, the crank angle sensor 17 is a means for detecting not only the crank angle but also the engine speed N.

The air flow meter 6, crank angle sensor 17 and O
The output signals from the x sensor 16 are both input to the control unit 30. Furthermore, control unit 3
The voltage of a buffer battery 20 is directly applied to the engine 0 via an engine key switch 21 as its operating power source and for detecting the power supply voltage.

Furthermore, the control unit 30 is provided with a water temperature sensor 22 for detecting the engine cooling water temperature as required. - Throttle sensor 23 including an idle switch that detects the throttle opening of the next throttle valve 7. Vehicle speed sensor 24 that detects vehicle speed. A signal from a neutral switch 25 or the like that detects the neutral position of the transmission is input. The control unit 30 performs arithmetic processing based on various input signals and outputs a drive pulse signal with an optimal pulse width to the fuel injection valve 11 to obtain a fuel injection amount for obtaining an optimal air-fuel ratio.

The control unit 30, as shown in FIG.
PU31. It has a P-ROM 32, an 0MO3-RAM 33, and an address decoder 34. Here, the RAM 33 is a rewritable storage means for learning control, and the RAM 33 is a rewritable storage means for learning control.
As the operating power source of 33, a battery 20 is connected via a suitable stabilized power source without using the engine key switch 21 in order to retain the memory contents even after the engine key switch 21 is turned off.

Among the input signals to the CPU 31, the air flow meter 6
, OR sensor 16. Battery 20. Since the voltage signals from the water temperature sensor 22 and the throttle sensor 23 are analog signals, they are inputted via an analog input interface 35 and an A/D converter 36. The A/D converter 36 is controlled by the CPU 31 via an address decoder 34 and an A/D conversion timing controller 37. The reference signal and position signal from the crank angle sensor 17 are inputted via a one-shot multi-circuit 38. The signal from the idle switch built in the throttle sensor 23 and the signal from the neutral switch 25 are input via the digital input interface 39, and the signals from the vehicle speed sensor 24 are inputted via the digital input interface 39.
The signal is inputted via the waveform shaping circuit 40.

The output signal (driving pulse signal for the fuel injection valve 11 ) from the CPU 31 is sent to the fuel injection valve 11 via the current waveform control circuit 41 .

Here, the CPU 31 performs input/output operations, arithmetic processing, etc. according to a program (stored in the ROM 32) based on the flowcharts (fuel injection amount calculation routine and learning subroutine) shown in FIGS. 4 and 5.
Controls fuel injection amount.

In addition, basic control amount (basic fuel injection amount) setting means.

Learning correction amount (coefficient) search means, feedback correction amount (
coefficient) setting means, control amount (fuel injection amount) calculation means, steady state detection means, learning correction amount (coefficient) correction means (including deviation peak value detection means, deviation average value calculation means, new learning correction amount calculation means) This function is achieved by the program.

Next, the operation will be explained with reference to the flowcharts of FIGS. 4 and 5.

In the fuel injection amount calculation routine shown in FIG.
(31 in the figure) shows the intake air flow rate Q obtained from the signal from the air flow meter 6 and the crank angle sensor 17.
The basic fuel injection amount Tp (=-Q/N) is calculated from the engine speed N obtained from the signal from the engine. This part corresponds to the basic control amount setting means.

In step 2, various correction coefficients C0EF are set as necessary.

In step 3, the learning correction coefficient K corresponding to the engine speed N representing the engine operating state and the basic fuel injection amount (load) Tp is determined.
Search for N. This part corresponds to the learning correction amount search means.

Here, the learning correction coefficient is set to 1 by dividing the map with the engine speed N on the horizontal axis and the basic fuel injection amount Tp on the vertical axis using a grid of about 8 x 8, dividing the area, and storing each area on the RAM 33. 1 is stored in the learning correction coefficient for each area. Note that all learning correction coefficients KN are set to an initial value of 1 at the time when learning has not started.

In step 4, a voltage correction numerator s is set based on the voltage value of the battery 20.

In step 5, it is determined whether the λ control condition is met.

Here, if the condition is not λ control condition, for example, high rotation, high load area, etc., the feedback correction coefficient α is clamped to the previous value (or reference value αI), and step 5
The process then proceeds to step 10, which will be described later.

In the case of the λ control condition, the output voltage V of the Ot sensor 16 is adjusted in steps 6 to 8. It compares t with slice level voltages V, .

This part corresponds to feedback correction amount setting means.

Specifically, in the case of integral control, when the comparison in step 6 determines that the air-fuel ratio = rich (V ot > V r.,), the feedback correction coefficient α is integrated with the previous value in step 7. (1) decrease, conversely, the air-fuel ratio -
When it is determined to be lean (Vow<Vrat), in step 8, the feedback correction coefficient α is increased by a predetermined integral (I) with respect to the previous value.

In the case of proportional-integral control, in addition to this, at the time of rich-lean reversal, a predetermined proportional amount (P) larger than the integral (I) is increased or decreased in the same direction as the integral (I) (see FIG. 6).

In the next step 9, the learning subroutine shown in FIG. 5 is executed. This will be discussed later.

Thereafter, in step 10, the fuel injection amount Ti is calculated according to the following equation. This part corresponds to the control amount calculation means.

T i =Tp −C0EF −KIl ・α+Ts
When the fuel injection amount Ti is calculated, a drive pulse signal having a pulse width of Ti is output at a predetermined timing in synchronization with the engine rotation, and is applied to the fuel injection valve 11 via the current waveform control circuit 41. Fuel injection takes place.

Next, the learning subroutine shown in FIG. 5 will be explained.

In step 11, it is determined whether the engine rotational speed N representing the engine operating state and the basic fuel injection amount 'rp are in the same range as the previous time. If it is the same area as the previous time, it is determined in step 12 whether flag F is set, and if it is not set, in step 13, the output of the O2 sensor 16 is reversed, that is, the direction of increase/decrease of the feedback correction coefficient α is changed. It is determined whether or not the reversal has occurred, and this flow is repeated. Each time the reversal is performed, the count value representing the number of reversals is incremented by 1 in step 14. When C=2, the process proceeds from step 15 to step 16 and the flag F is set. Set. This flag F is set when the output of the 02 sensor 16 is reversed twice in the same area.

After setting the flag F, if it is determined in step 11 that the area is the same as the previous time, the process proceeds to step 17 via step 12. Steps 11 to 16 correspond to the steady state detection means, and are: - The engine operating state is in one of the divided regions; - The feedback correction coefficient α is in the increasing or decreasing direction or reversed a predetermined number of times (twice) or more. It is detected that a steady state is reached.

In a steady state, it is determined in step 17 whether the output of the 0□ sensor 16 has been reversed, that is, the direction of increase/decrease of the feedback correction coefficient α has been reversed, and when this flow is repeated and reversed, it is determined to be steady in step 18. Therefore, it is determined whether or not it is the third inversion in the same area. If it is the third inversion, in step 19, the reference value α of the current feedback correction coefficient α is set.

The deviation Δα (=α−αI) is temporarily stored as Δα1. Thereafter, when the fourth reversal is detected, the process proceeds to steps 20 to 24 and learning is performed based on data from the third to fourth reversal (see FIG. 6). The same goes for steps 20 to 2 when the fifth or more reversal is detected.
Proceed to step 4 and perform learning based on the data from the previous reversal to the current reversal.

At the time of the fourth or more reversal, in step 20, the deviation Δα (==α−
α1) is temporarily stored as Δα2. The Δα and Δα stored at this time are the upper and lower peak values of Δα from the previous (for example, the third) reversal to the current (for example, the fourth) reversal, as shown in FIG. Here, step 1
The portions 7 to 20 correspond to the deviation peak value detection means of the learning correction amount correction means.

These upper and lower peak values Δα1. Deviation Δ based on Δα2
Since the average value τi of α can be calculated, in step 21, the average value τi of the deviation Δα is calculated based on the following equation. This step 21 corresponds to the deviation average value calculating means of the learning correction amount correcting means.

τz=(Δα1 +Δα/2) Next, in step 22, a stored learning correction coefficient KIl corresponding to the current area is retrieved. However, in reality, the one retrieved in step 3 may be used.

Next, in step 23, the current learning correction coefficient K is calculated according to the following formula.
A new learning correction coefficient Kl0w is calculated by adding a predetermined proportion of the average value of the deviation Δα (=α−α,) of the feedback correction coefficient α from the reference value α1 to N,
Correct and rewrite the learning correction coefficient data in the same area.

This step 23 corresponds to a new learning correction amount calculation means of the learning correction amount correction means.

K j' (, law)←KA+τff/M (M is a constant, M>1) After this, in step 24, the value of Δα2 is substituted into Δα1 for the next calculation.

If it is determined in step 11 that the engine operating state is no longer in the same range as the previous time, the count value C is cleared and the flag F is reset in step 25.

Although the air-fuel ratio learning control device has been described above, the present invention can of course be applied to an idle rotation speed learning control device.

Specifically, in the idle speed learning control device, the basic control amount ISCtw is set from the water temperature, and the learning correction amount l is set from the water temperature.
5C1e, set the feedback correction amount l5Cfb by comparing the target idle rotation speed Ns set from the water temperature and the actual idle rotation speed, and calculate the control amount l5Cdy to the idle control valve 10 by the following (11 formula), In a steady state, learning is performed according to the following formula (2), but in this case, the average value - of the deviation Δl5Cfb of the feedback correction amount l5Cfb from the reference value is calculated using the same method as described above, Based on this, a new learning correction amount l
5C1e should be set.

I5Cdy=ISCtw+I5C1e+I
5Cfb-(1) IS Cle (lll1w) ←
I S Cle + τI S Cfb/M (2) The idle control valve 10 is equipped with a valve opening coil and a valve closing coil, and pulse signals are sent to these coils in a mutually determined state, and the pulse signals are The opening degree is adjusted according to the duty ratio of the valve opening coil, and the l5Cdy mentioned above is the time percentage (%) that the valve opening coil is ON.

<Effects of the Invention> As explained above, according to the present invention, each time the direction of increase/decrease in the feedback correction amount is reversed, the peak value of the deviation from the reference value of the feedback correction amount at that time is captured, and the two before and after are detected.
Since the average value is calculated from the upper and lower peak values obtained by the detection of times, and learning is performed based on this, it is possible to achieve the effect that high-accuracy and high-speed learning is possible.

[Brief explanation of the drawing]

Fig. 1 is a functional block diagram showing the configuration of the present invention, Fig. 2 is a block diagram showing an embodiment of the invention, Fig. 3 is a block circuit diagram of the control unit in Fig. 2, and Figs. 5
The figure is a flowchart showing control details, and FIG. 6 is a control characteristic diagram.

Claims (1)

[Scope of Claims] A basic control amount setting means for setting a basic control amount corresponding to a control target value of a controlled object of the internal combustion engine such as an air-fuel ratio or an idle rotation speed; a rewritable storage means that stores a learning correction amount for correcting the basic control amount for each region of the engine operating state; and a learning correction amount for the corresponding region is searched from the storage means based on the actual engine operating state. a learning correction amount search means for comparing a control target value and an actual value, and a feedback unit for increasing or decreasing a feedback correction amount by a predetermined amount to correct the basic control amount so that the actual value approaches the control target value. A control amount is determined from the correction amount setting means, the basic control amount set by the basic control amount setting means, the learning correction amount searched by the learning correction amount searching means, and the feedback correction amount set by the feedback correction amount setting means. a control amount calculation means for calculating the control amount; a control means for operating according to the control amount to control control targets of the internal combustion engine such as an air-fuel ratio and an idle speed; and an actual engine operating state is in a steady state. Steady state detection means for detecting that the steady state is being detected, and the device learns the deviation of the feedback correction amount from the reference value during the steady state detection and corrects the learning correction amount corresponding to the region of the engine operating state during that period in a direction to reduce this deviation. A learning control device for an internal combustion engine, comprising: a learning correction amount correcting means for rewriting the learning correction amount, and the learning correction amount correcting means changes the current feedback correction amount each time the direction of increase/decrease of the feedback correction amount is reversed during steady state detection. deviation peak value detection means for detecting the peak value of the deviation from the reference value; and a deviation average for calculating the average value from the upper and lower peak values of the deviation obtained by the previous and the previous two detections by the deviation peak value detection means. a new learning correction amount that calculates a new learning correction amount by adding a predetermined percentage of the average value of the deviation to the current learning correction amount every time the average value of the deviation is obtained by the deviation average value calculation means; A learning control device for an internal combustion engine, characterized in that it is configured to include a calculation means;