JP2715208B2 - Air-fuel ratio learning control device for internal combustion engine - Google Patents

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 learning control system for an internal combustion engine, and more particularly, to a technology for improving air-fuel ratio controllability in a low exhaust temperature region where an air-fuel ratio sensor provided in an engine exhaust system is inactive. About.

[0002]

2. Description of the Related Art Conventionally, in an electronically controlled fuel injection device for an internal combustion engine having an air-fuel ratio feedback correction control function,
JP-A-60-90944, JP-A-61-1901
As disclosed in Japanese Patent Application Publication No. 42-42, etc., there is one that adopts air-fuel ratio learning control. The air-fuel ratio feedback correction control is an oxygen sensor (air-fuel ratio sensor) whose output value changes in response to the rich / lean actual air-fuel ratio with respect to a target air-fuel ratio (for example, stoichiometric air-fuel ratio) in response to the oxygen concentration in the engine exhaust.
And an air-fuel ratio feedback correction coefficient LMD (air-fuel ratio feedback control value) is set by proportional / integral control or the like based on the result of the determination. Then, the actual air-fuel ratio is corrected to the target air-fuel ratio by correcting the basic fuel injection amount Tp calculated from the detected value of the intake air flow rate or the suction negative pressure and the engine rotation speed by the air-fuel ratio feedback correction coefficient LMD. This is for feedback control.

Here, the deviation of the air-fuel ratio feedback correction coefficient LMD from the reference value (target convergence value) is learned for each of a plurality of operating regions, and the air-fuel ratio learning correction coefficient KB
LRC (air-fuel ratio learning control value) is determined, and the basic fuel injection amount Tp is corrected by the learning correction coefficient KBLRC so that the base air-fuel ratio obtained without the feedback correction coefficient LMD substantially matches the target air-fuel ratio. During the fuel ratio feedback control, the fuel injection amount Ti is further corrected by the correction coefficient LMD.
Is calculated.

[0004] This makes it possible to perform fuel correction corresponding to different correction requests for each operating condition, stabilize the actual air-fuel ratio near the stoichiometric air-fuel ratio, maintain good conversion efficiency in the three-way catalytic converter, and improve the exhaust gas. Harmful component concentration can be controlled to a low level.

[0005]

Incidentally, an oxygen sensor (air-fuel ratio sensor) provided in an engine exhaust system for detecting the air-fuel ratio of the engine intake air-fuel mixture via the oxygen concentration in the exhaust gas.
In the inactive state where the device temperature is low, the desired output characteristics cannot be obtained, and the device temperature must be kept high to obtain the desired output characteristics. However, when a small displacement engine or a heater for heating the element is not provided in the oxygen sensor, it becomes difficult to activate the oxygen sensor in the entire operation range, and the operation range is relatively low in the exhaust gas temperature ( In the low-load low-speed range), the oxygen sensor becomes inactive and the desired air-fuel ratio feedback control cannot be performed, and the air-fuel ratio learning may not proceed. As described above, if the air-fuel ratio learning cannot be advanced in the low exhaust temperature region, it is not possible to compensate for the deviation of the base air-fuel ratio due to component deterioration or the like in the operation region, and the operability and exhaust characteristics are reduced. There is a problem that the fuel ratio learning control makes it impossible to maintain the fuel ratio satisfactorily.

Conventionally, an air-fuel ratio learning value in an unlearned region is estimated and rewritten based on an air-fuel ratio learning value in a learned region (hereinafter, such learning is referred to as estimation learning). There is an air-fuel ratio learning device configured to be able to avoid deterioration of the fuel ratio controllability. However, in this case, the oxygen sensor is not divided into regions based on activation / inactivation, and the oxygen sensor does not exhibit the intended output characteristics in the inactive region, so that highly accurate air-fuel ratio learning may not be performed. Nevertheless, the learning value of the same inactive region is estimated based on the uncertain learning result in the inactive region, or the uncertain learning result in the inactive region is estimated as the learning value of the active region. There is a problem that high-precision estimation learning cannot be performed.

The present invention has been made in view of the above problems, and even if the air-fuel ratio sensor becomes inactive in a low exhaust gas temperature range and the air-fuel ratio learning control becomes impossible, the base air in such an operation region is not controlled. It is an object of the present invention to provide a device capable of performing air-fuel ratio learning control capable of compensating for a difference in fuel ratio with high accuracy.

[0008]

Therefore, an air-fuel ratio learning control device for an internal combustion engine according to the present invention is configured as shown in FIG. In FIG. 1, an air-fuel ratio sensor changes its output value in response to the concentration of a specific component in exhaust gas that changes according to the air-fuel ratio of an engine intake air-fuel mixture. The air-fuel ratio detected by the above is compared with the target air-fuel ratio, and the air-fuel ratio feedback control value of the engine is set so that the actual air-fuel ratio approaches the target air-fuel ratio.

On the other hand, the storage means rewritably stores the air-fuel ratio learning control value for each of a plurality of operating regions based on the engine operating conditions, and the air-fuel ratio learning means stores the target convergence value of the air-fuel ratio feedback control value. , And the air-fuel ratio learning control value stored in the storage means corresponding to the corresponding operating region is corrected and rewritten in a direction to reduce the deviation.

[0010] The low exhaust temperature region estimating learning means, when the predetermined low exhaust temperature region preset in the storage means has not been learned, stores the specified operating region other than the predetermined low exhaust temperature region. A learned air-fuel ratio learning control value correspondingly stored, and the air-fuel ratio learning in the air-fuel ratio enrichment direction
The air-fuel ratio learning control value corresponding to the predetermined low exhaust gas temperature range is rewritten based only on the learning control value. The air-fuel ratio control means controls the air-fuel ratio of the engine intake air-fuel mixture based on the air-fuel ratio feedback control value and the air-fuel ratio learning control value stored in the operating area in the storage means.

[0011]

[0012]

With this configuration, when the predetermined low exhaust gas temperature region is not yet learned, the learned air-fuel ratio learning stored in correspondence with the specified operation region other than the predetermined low exhaust gas temperature region. Based on the control value, the learning control value in the predetermined low exhaust temperature range is rewritten. Therefore, even when the air-fuel ratio learning does not proceed due to the inactivity of the air-fuel ratio sensor within the low exhaust temperature range, the air-fuel ratio learning control is not performed at all and the air-fuel ratio learning control is not performed and the initial state is not left. In response to a learning correction request in a low exhaust temperature region, the learning result in an operating region other than the temperature region, in other words, in an operating region where the exhaust gas temperature is relatively high and the air-fuel ratio sensor can be activated can be learned. Correction can be performed, and a delay in learning progress due to inactivation of the air-fuel ratio sensor can be compensated.

Further , when the air-fuel ratio learning control value in the low exhaust temperature region is rewritten based on the air-fuel ratio learning control value learned in a region other than the low exhaust temperature region, only the air-fuel ratio learning control value in the air-fuel ratio enrichment direction is used. the as used, it is not performed to estimate the air-fuel ratio learning control value in the low exhaust gas temperature region based on the air-fuel ratio learning correction value of the lean direction
Thus, it is possible to avoid that the result of the erroneous leaning direction learning performed outside the low exhaust temperature region is reflected in the low exhaust temperature region.

[0014]

Embodiments of the present invention will be described below. In FIG. 2 showing one embodiment, air is sucked into an internal combustion engine 1 from an air cleaner 2 through an intake duct 3, a throttle valve 4 and an intake manifold 5. In each branch of the intake manifold 5, a fuel injection valve 6 is provided for each cylinder. The fuel injection valve 6 is an electromagnetic fuel injection valve that is energized by a solenoid and opens, and is deenergized and closed by being energized by a drive pulse signal from a control unit 12, which will be described later. Fuel which is pressure-fed from a fuel pump (not shown) and adjusted to a predetermined pressure by a pressure regulator is intermittently injected and supplied to the engine 1.

An ignition plug 7 is provided in each combustion chamber of the engine 1 to ignite a mixture by spark ignition. Then, exhaust gas is discharged from the engine 1 through the exhaust manifold 8, the exhaust duct 9, the three-way catalyst 10, and the muffler 11. The control unit 12 includes a CPU, RO
A microcomputer including an M, a RAM, an A / D converter, an input / output interface, and the like is provided. The microcomputer receives input signals from various sensors, performs arithmetic processing as described later, and operates the fuel injection valve 6. Control.

The various sensors include an intake duct 3
An air flow meter 13 is provided therein, and outputs a signal corresponding to the intake air flow rate Q of the engine 1. Further, a crank angle sensor 14 is provided. For example, in the case of a four-cylinder engine, a reference signal REF for every 180 ° crank angle and a crank angle of 1 °
Alternatively, a unit signal POS for every 2 ° is output. Here, the engine speed Ne can be calculated by measuring the period of the reference signal REF or the number of occurrences of the unit signal POS within a predetermined time.

Further, a water temperature sensor 15 for detecting a cooling water temperature Tw of the water jacket of the engine 1 is provided.
Further, an oxygen sensor 16 as an air-fuel ratio sensor is provided at the collecting portion of the exhaust manifold 8, and detects the air-fuel ratio of the intake air-fuel mixture through the oxygen concentration in the exhaust gas. The oxygen sensor 16
Is an oxygen concentration sensor whose output value changes in response to the concentration of oxygen, which is a specific component in the exhaust gas. The oxygen concentration in the exhaust gas suddenly changes at a stoichiometric air-fuel ratio (the target air-fuel ratio in this embodiment). Utilizing this fact, it is known to detect rich / lean of the actual air-fuel ratio with respect to the stoichiometric air-fuel ratio.

Here, the CPU of the microcomputer built in the control unit 12 performs arithmetic processing according to the programs on the ROM shown in the flowcharts of FIGS. The fuel injection amount Ti is set while executing the fuel ratio learning correction control, and the fuel supply to the engine 1 is controlled.

In this embodiment, the functions of the air-fuel ratio feedback control means, the air-fuel ratio learning means, the air-fuel ratio control means, and the low exhaust temperature region estimation learning means are as shown in the flowcharts of FIGS. The control unit 12 is provided as software, and the storage means corresponds to a RAM with a backup function of a microcomputer (not shown) built in the control unit 12.

The program shown in the flowchart of FIG. 3 is based on a basic fuel injection amount Tp = K × Q / N (K is a constant) calculated based on the intake air flow rate Q and the engine speed Ne.
Is a program for setting an air-fuel ratio feedback correction coefficient LMD (air-fuel ratio feedback control value) by the proportional / integral control, which is executed for each revolution (1 rev) of the engine 1.

First, in step 1 (referred to as S1 in the figure, the same applies hereinafter), a voltage signal output from the oxygen sensor 16 according to the oxygen concentration in the exhaust gas is read. In the next step 2, it is determined whether or not conditions for performing the air-fuel ratio feedback control are satisfied. Conditions for stopping the air-fuel ratio feedback control include, for example, starting, low water temperature,
When the exhaust gas temperature is low and the oxygen sensor 16 is in an inactive state, such as during idling, abnormal output of the oxygen sensor 16, or abnormal control cycle, the feedback control is stopped as an abnormal output because the output fluctuation width is small. It has become so.

When it is determined in step 2 that the air-fuel ratio feedback control condition is satisfied, the process proceeds to step 3, in which the voltage signal from the oxygen sensor 16 read in step 1 corresponds to the stoichiometric air-fuel ratio (target air-fuel ratio). With the slice level (for example, 500 mV). When it is determined that the voltage signal from the oxygen sensor 16 is higher than the slice level and the air-fuel ratio is richer than the stoichiometric air-fuel ratio, the process proceeds to step 4, where it is determined whether the current rich determination is the first time.

When the rich determination is performed for the first time, the routine proceeds to step 5, where the air-fuel ratio feedback correction coefficient LMD set up to the previous time is set to the maximum value a. In the next step 6, control for reducing the correction coefficient LMD is performed by subtracting a predetermined proportionality constant P from the correction coefficient LMD up to the previous time, and the fuel injection amount is reduced by the control for reducing the correction coefficient LMD to reduce the air-fuel ratio. Correct in the direction. In step 7, a flag FP indicating that the proportional control has been executed is set.
Is set to 1.

On the other hand, if it is determined in step 4 that the rich determination is not the first time, the process proceeds to step 8, in which a value obtained by multiplying a predetermined integral constant I by the latest fuel injection amount Ti is calculated from the correction coefficient LMD up to the previous time. The subtraction is performed to update the correction coefficient LMD. When it is determined in step 3 that the air-fuel ratio is lean with respect to the target, similarly to the rich determination, first, in step 9, it is determined whether or not the current lean determination is the first time. If it is the first time, the process proceeds to step 10 and the correction coefficient LMD up to the previous time is set to the minimum value b.

In the next step 11, the fuel injection amount is increased by adding the proportionality constant P to the correction coefficient LMD up to the previous time to correct the air-fuel ratio in the rich direction. In step 12, the flag FP is set to 1 Is set. If it is determined in step 9 that the lean determination is not the first time,
Proceeding to 13, the value obtained by multiplying the integration constant I by the latest fuel injection amount Ti is added to the correction coefficient LMD up to the previous time.

In this way, the air-fuel ratio feedback correction coefficient LMD is proportionally and integrated controlled so that the actual air-fuel ratio detected by the oxygen sensor 16 approaches the target air-fuel ratio. In the present embodiment, the air-fuel ratio feedback correction coefficient LMD is set by proportional / integral control. However, only integral control or proportional / integral / differential control may be used.

The program shown in the flow chart of FIG. 4 is an air-fuel ratio learning program for each operating region, and is executed every predetermined minute time (for example, 10 ms). In step 21, the flag FP is determined. When FP is 1, the process proceeds to step 22 to reset the FP to zero, and then performs various processes according to the present program. When it is zero, the program is terminated as it is.

When the FP is reset to zero in step 22,
In the next step 23, the basic fuel injection amount Tp (= K × Q / N; K is a constant) representing the engine load and the engine speed N
On the learning map (see FIG. 6) in which the air-fuel ratio learning correction coefficient KBLRC (air-fuel ratio learning control value) is rewritably stored for each of the operating regions divided into a plurality by using the engine operating condition of e as a parameter, In order to identify the corresponding area, the latest basic fuel injection amount Tp and the engine speed Ne are read respectively.

Then, in the next step 24, step 23
The basic fuel injection amount Tp and the engine speed Ne read by
Are stored corresponding to the area on the learning map corresponding to
The air-fuel ratio learning correction coefficient KBLRC is read and
RC_OLDSet to. In step 25, the air-fuel ratio filter
Average of maximum and minimum values a and b of feedback correction coefficient LMD
(= (A + b) / 2) and the target convergence value (of the correction coefficient LMD)
This is an initial value, and in this embodiment, a predetermined ratio of deviation from 1.0)
X is the air-fuel ratio learning correction coefficient KBLRC _OLDValue added to
To the new air-fuel ratio learning correction coefficient KBLRC
_NEW(← KBLRC_OLD+ X · {(a + b) /2-1.0})
Set as

By this learning, the correction amount based on the air-fuel ratio feedback correction coefficient LMD (air-fuel ratio feedback control value) is converted into the air-fuel ratio learning correction coefficient KBLRC (air-fuel ratio learning control value) for each operating region, and the air-fuel ratio feedback correction coefficient is obtained. The deviation between the LMD and the target convergence value can be reduced, and the air-fuel ratio feedback correction coefficient LMD can be stabilized near the target convergence value, and can cope with different correction requests depending on the operation region.

In step 26, the air-fuel ratio learning correction coefficient
KBLRC _NEW is rewritten as map data as update data corresponding to the corresponding area on the learning map. Furthermore,
In the next step 27, the learning counter for each area corresponding to the area in which the map data has been rewritten in step 26 is incremented so that the number of learning experiences for each operating area can be determined for each operating area.

The final fuel injection amount Ti is calculated as follows. That is, while calculating the basic fuel injection amount Tp from the intake air flow rate Q and the engine rotation speed Ne, various correction coefficients CO are set according to operating conditions such as the cooling water temperature Tw, and the air-fuel ratio feedback correction coefficient LM is further set.
D and the air-fuel ratio learning correction coefficient KBLRC stored in the corresponding operation area on the learning map, and Ti = T
The calculation is performed as p × CO × LMD × KBLRC. Then, at a predetermined injection timing synchronized with the engine rotation, a drive pulse signal having a pulse width corresponding to the latest calculated fuel injection amount Ti is output to the fuel injection valve 6, thereby controlling the fuel supply to the engine. Is done.

After incrementing the learning counter in step 27, in step 28, the operating range on the learning map on which the desired air-fuel ratio cannot be learned because the exhaust gas temperature is low, the oxygen sensor 16 becomes inactive. For a (predetermined low exhaust gas temperature region), estimation learning that reflects the learned result in the active region is performed. The details of the estimation learning in step 28 are shown in the flowchart of FIG.

In the flowchart of FIG. 5, first, at step 31, it is determined whether or not the current driving condition is included in a presumed learning area predetermined on the learning map. As shown in FIG. 6, the estimation learning region includes a predetermined low exhaust temperature region (inactive region) in a low load low rotation region where the possibility that the oxygen sensor 16 becomes inactive due to a low exhaust temperature is large. And an active region in which the oxygen sensor 16 is predicted to be activated according to a relatively high exhaust temperature, which is a specific operating region adjacent to the predetermined low exhaust temperature region.

The inactive region is a region in which the air-fuel ratio feedback control cannot be performed satisfactorily because the oxygen sensor 16 is inactive, thereby preventing the progress of the air-fuel ratio learning. Are all expected to be able to activate the oxygen sensor 16, but in the present embodiment, the active region used for estimation learning described later is adjacent to the inactive region, and the required correction level is predicted to be relatively close. In order to limit to an area, an estimation learning area is set as shown in FIG. 6, and an active area for estimation learning is specified.

When it is determined in step 31 that the vehicle is in the estimated learning region, the process proceeds to step 32, in which it is determined whether or not the active region in the estimated learning region has been learned. A determination is made based on whether or not a correspondingly set learning counter is equal to or greater than a predetermined value. here,
The determination as to whether or not the active region has been learned may be made on condition that all the learning operation regions included in the active region have been learned. The condition may be that at least a predetermined number (including 1) has been learned.

When it is determined that the active region has been learned, the routine proceeds to step 33, where the average value of the air-fuel ratio learning correction coefficient KBLRC in each learned region in the active region is 1.
It is determined whether the value exceeds 0 and the correction in the enrichment direction has been learned, or conversely, is less than 1.0 and the correction in the leaning direction has been learned. In the air-fuel ratio learning described above, if the air-fuel ratio learning is performed when the canister purge is performed, the air-fuel ratio learning correction coefficient KBLRC is set to 1.0 in an attempt to compensate for the temporary air-fuel ratio enrichment due to the canister purge. Will be erroneously learned to a value less than. Therefore, the average of the learning results in the active area is 1.0
If the value is lower than the predetermined value, the learning may be performed in response to the fact that the base air-fuel ratio has been enriched, but the canister purge may have been erroneously learned.

As in the present embodiment, when estimation learning is performed in which the result of the learned region is reflected in the unlearned region, the result of the erroneous learning based on the canister purge (correction of the leaning direction) is not learned. If this is reflected in the area, if the request for the area to be estimated is in the direction of enrichment, the step of the actual correction level for the request is further increased by estimation learning, which is not preferable.

Therefore, when there is a learning result in the enrichment direction that may cause erroneous learning due to the canister purge,
That is, when it is determined in step 33 that the learning average value is less than 1.0, the learning result may be correct,
In order to prevent the result of the erroneous learning by the canister purge from being reflected in the unlearned area, the present program is terminated without performing the estimation learning.

On the other hand, if it is determined in step 33 that the learned average value exceeds 1.0, it is considered that at least the canister purge has not been erroneously learned in the active region, and the routine proceeds to the next step 34. In step 34, it is determined whether or not the inactive area has been learned. Here, when it is determined by the learning counter that all of the plurality of learning driving regions included in the non-active region have not been learned, it is determined as unlearned in the non-active region.

If it is determined in step 34 that all the non-active areas are unlearned, the process proceeds to step 35, in which a process of reflecting the learning result in the active area in the estimated learning area to the non-learned non-active area is performed. Do. Specifically, the average value of the learned air-fuel ratio learning control values in the active region included in the estimated learning region is updated and stored as the learning values of all the operating regions in the inactive region. In this case, it is assumed that the greater the number of learned regions in the active region, the better the accuracy of estimation learning is, and that all operating regions in the inactive region are unlearned.

As described above, the learning result in the adjacent active region is reflected in the inactive region in which the air-fuel ratio learning cannot be performed because the oxygen sensor 16 becomes inactive. In the active region, it is possible to perform the air-fuel ratio learning correction substantially corresponding to the demand, and it is possible to improve the operability and the exhaust characteristics in the inactive region. In the above embodiment, the learning average value in the specified active region adjacent to the non-active region is estimated as the learning value in the non-active region. There is a method of performing estimation learning by specifying a direction.

That is, estimation learning is performed between operation regions on the learning map in which any one of the basic fuel injection amount Tp, the engine rotation speed Ne, and the intake air flow rate Q is at substantially the same level. Is used as the estimation learning condition, when the driving region in the inactive region at substantially the same Tp level as the learned region in the active region is unlearned, the estimation learning from the active region to the inactive region is performed. Let them do it.

Therefore, in this case, the equal Tp
When all the regions located above are unlearned and have been learned in the specified active region located on the same extension of the equal Tp, estimation learning is performed, and the inactive region is The estimation learning is not performed at once, but is performed for each operation region group of equal Tp. However, in this case, in each operation region where the predetermined operation condition is substantially the same level, the estimation learning is performed assuming that the correction requests are substantially the same, so that the active region is adjacent to the inactive region as shown in FIG. It is not necessary to limit to an area, and an area other than the inactive area may be regarded as an active area.

Further, the method of estimation learning may be other than the above, but it is a necessary condition that the inactive operation region to be estimated and learned is not learned. This is because, even in the inactive region, the desired air-fuel ratio learning may be performed by activating the oxygen sensor 16 depending on conditions, and it is desirable to give priority to the actual learning result over the estimation learning result. That's why.

In addition, the learning result in the inactive area is
It is not reflected on the unlearned area in the active area. This is because there is a possibility that the learning result in the inactive operation region is obtained in a state where the oxygen sensor 16 is not sufficiently activated, and in this case, it is possible to perform highly accurate estimation learning. This is because they cannot.

[0047]

As described above, according to the present invention, the air-fuel ratio sensor becomes inactive due to the low exhaust gas temperature, and the air-fuel ratio sensor cannot perform the desired air-fuel ratio learning (low exhaust gas temperature region). Since the learning result in the region where the fuel ratio sensor is activated is reflected, it is possible to compensate for the deviation of the base air-fuel ratio in the inactive region (low exhaust temperature region), and to improve the drivability in this region. ， Improve exhaust properties.

Further , as described above, when the learning result of the active region is reflected in the inactive region (low exhaust temperature region), only the air-fuel ratio learning result in the air-fuel ratio enrichment direction is used. Erroneous learning in the leaning direction due to is set as the learning value of the inactive region.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of the present invention.

FIG. 2 is a system schematic diagram showing one embodiment of the present invention.

FIG. 3 is a flowchart showing air-fuel ratio feedback control.

FIG. 4 is a flowchart showing air-fuel ratio learning control.

FIG. 5 is a flowchart showing estimation learning for an inactive area.

FIG. 6 is a diagram showing a state of a learning map in the embodiment.

[Explanation of symbols]

 1 engine 6 fuel injection valve 12 control unit 13 air flow meter 14 crank angle sensor 16 oxygen sensor (air-fuel ratio sensor)

Claims (1)

(57) [Claims] An air-fuel ratio sensor whose output value changes in response to the concentration of a specific component in exhaust gas that changes according to the air-fuel ratio of an engine intake air-fuel mixture, an air-fuel ratio detected by the air-fuel ratio sensor, and a target air-fuel ratio Air-fuel ratio feedback control means for setting the air-fuel ratio feedback control value of the engine such that the actual air-fuel ratio approaches the target air-fuel ratio by comparing the air-fuel ratio with the target air-fuel ratio. Storage means for rewritably storing a fuel ratio learning control value; learning of a deviation of the air-fuel ratio feedback control value from a target convergence value; and storing the air-fuel ratio stored in the storage means corresponding to a corresponding operating region. An air-fuel ratio learning unit that corrects and rewrites the learning control value in a direction to reduce the deviation; and when a predetermined low exhaust temperature region preset in the storage unit is unlearned, The learned air-fuel ratio learning control value stored corresponding to the specified operation region other than the predetermined low exhaust gas temperature region, and the air-fuel ratio enrichment direction
A low exhaust temperature region estimating learning unit that rewrites an air-fuel ratio learning control value corresponding to the predetermined low exhaust temperature region based only on the air-fuel ratio learning control value of the air-fuel ratio feedback control value and the storage unit. An air-fuel ratio learning control device for an internal combustion engine, comprising: air-fuel ratio control means for controlling an air-fuel ratio of an engine intake air-fuel mixture based on an air-fuel ratio learning control value stored in a region.