JP2640566B2 - Air-fuel ratio learning control device for internal combustion engine - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION <Industrial Application Field> The present invention relates to an air-fuel ratio learning control device for an internal combustion engine, and more specifically, to respond to a more detailed air-fuel ratio correction request for each operating region while saving memory capacity. To a device that makes it possible.

<Prior Art> Conventionally, an internal combustion engine equipped with an electronically controlled fuel injection device having an air-fuel ratio feedback correction control function has been disclosed in
As disclosed in JP-A-90944 and JP-A-61-190142, learning control of an air-fuel ratio is employed.

In the air-fuel ratio feedback correction control, a basic fuel injection amount Tp calculated from parameters of an engine operating state (for example, intake air flow rate Q and engine speed N) related to the amount of air taken into the engine is provided in the engine exhaust system. The actual air-fuel ratio is corrected by the air-fuel ratio feedback correction coefficient LMD set by proportional / integral control based on the rich / lean relative to the target air-fuel ratio (for example, the stoichiometric air-fuel ratio) determined by the oxygen sensor. The feedback control is performed on the air-fuel ratio.

Here, a deviation of the air-fuel ratio feedback correction coefficient LMD from a reference value (a value that does not substantially correct the fuel amount and becomes a target convergence value) is learned for each of a plurality of predetermined operating regions. To determine the learning correction coefficient KBLRC, the basic fuel injection amount Tp is corrected by the learning correction coefficient KBLRC,
The base air-fuel ratio obtained without the correction coefficient LMD is made substantially equal to the target air-fuel ratio, and during the air-fuel ratio feedback control, the fuel injection amount Ti is further corrected by the correction coefficient LMD.

This makes it possible to perform a correction corresponding to the required value of the air-fuel ratio correction that differs for each operating condition, and to provide an air-fuel ratio feedback correction coefficient
By stabilizing the LMD near the reference value, it is possible to improve the air-fuel ratio controllability especially in an excessive state.

<Problems to be Solved by the Invention> By the way, as described above, the learning correction coefficient KB
When the LRC is set so that the air-fuel ratio correction corresponding to the difference in the correction request depending on the operating condition can be performed, the operation area covered by one learning correction coefficient KBLRC by making the divided operation area in the learning map as small as possible It is desirable to reduce the area. However, if the operating area is finely divided and the learning correction coefficient KBLRC is learned and stored for each area, the required memory capacity for storing the learning correction coefficient KBLRC becomes large, and each line There is a problem in that the learning opportunity for each area decreases and the progress of learning slows down.

Therefore, conventionally, more emphasis has been placed on saving the memory capacity and securing the learning progress speed, and it has been difficult to perform the learning by sufficiently dividing the operation region into sufficiently small areas.

Here, a specific example of a problem that occurs when the learning region is relatively large is described. For example, as shown in FIG. 13, an air-fuel ratio change such that the air-fuel ratio becomes richer as the rotation speed increases. In the case where deterioration of a component having characteristics occurs, when the rotation speed slowly increases with a constant boost as shown in FIG. 12, the following problems may occur.

That is, as shown in FIG. 12, when the boost (basic fuel injection amount) is constant and the rotation speed changes slowly,
Since the characteristics of the air-fuel ratio change are as shown in the figure, it is necessary to increase the lean correction amount as the rotation speed increases.
However, for example, the learning correction coefficient KB is determined for each of a plurality of operating regions based on the basic fuel injection amount Tp and the engine speed N.
When learning the LRC, since there is a range of rotation speed in one operation area, the largest leaning request correction amount on the high rotation side just before passing through the corresponding area is learned and stored corresponding to each learning area. Will be.

Therefore, when the same region is again entered from the low rotation side after such learning is performed, as shown in FIG.
If the leaning correction by the learning correction coefficient KBLRC becomes excessive and the base air-fuel ratio becomes lean, the overall base air-fuel ratio tends to lean, and conversely, if learning is performed in a state where the rotation is slowly reduced, In each of the learning regions, the smallest leaning correction amount in the learning region is learned and stored. Therefore, when the same operation is repeated again, the average base air-fuel ratio is enriched in reverse. It will be.

That is, since the segmented operation region for learning the learning correction coefficient KBLRC has a certain size, especially when the operation condition changes slowly, a learning value corresponding to the operation condition immediately before exiting the corresponding region is applied to the entire region. Since it is stored as the applied learning value, the larger the step of the required correction amount in one learning area is,
This is to deteriorate the air-fuel ratio controllability.

Such a problem can be solved by sufficiently dividing the operation region so as to reduce the change of the required correction value in one divided region. However, in consideration of the memory capacity and the learning progress direction as described above, it is sufficient. Was unable to perform the learning by subdividing the operating region into smaller ones.

The present invention has been made in view of the above-described problems, and performs air-fuel ratio control with high accuracy in response to a difference in correction request due to a difference in operating conditions, and can save a memory capacity. It is another object of the present invention to provide an air-fuel ratio learning control device capable of ensuring a learning progress speed.

<Means for Solving the Problems> 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 operating condition detecting means detects an engine operating condition including at least an operating parameter related to an amount of air taken into the engine, and a basic fuel supply amount setting means sets a basic fuel supply amount based on the engine operating condition. Set the supply amount.

Further, the air-fuel ratio detecting means detects the air-fuel ratio of the engine intake air-fuel mixture, and the air-fuel ratio feedback correction value setting means compares the detected actual air-fuel ratio with the target air-fuel ratio to determine the actual air-fuel ratio. Is set to an air-fuel ratio feedback correction value for correcting the basic fuel supply amount so as to approach the target air-fuel ratio.

On the other hand, the learning correction value area-specific storage means is configured to rewritably store an air-fuel ratio learning correction value for correcting the basic fuel supply amount for each of a plurality of operating areas based on engine operating conditions, The learning correction means learns the deviation of the air-fuel ratio feedback correction value from the target convergence value, and learns the air-fuel ratio of the corresponding segmented operation area stored in the learning correction value area-specific storage means in a direction to decrease the deviation. Correct and rewrite the correction value. Further, the sub-region storage unit further divides only the operating region corresponding to the learning correction value region storage unit into a plurality of sub-regions, and stores the air-fuel ratio learning correction value corrected by the learning value correction unit. ,
The sub-region is temporarily stored as a sub-division learning correction value in association with a corresponding sub-region of the plurality of sub-regions. And
The correction unit based on the average value, when a corresponding one of the divided operation regions in the plurality of divided operation regions in the learning correction value region storage unit changes, for each of the subdivision regions stored in the subdivision region storage unit. An average value of the plurality of subdivision learning correction values is obtained, and based on the average value, the air-fuel ratio learning correction value of the corresponding divided operation region in the learning correction value region-specific storage means is rewritten and corrected.

Here, the fuel supply amount setting means determines the final fuel supply amount based on the basic fuel supply amount, the air-fuel ratio feedback correction value, and the air-fuel ratio learning correction value of the corresponding divided operation area stored in the learning correction value area storage means. The fuel supply amount is set, and the fuel supply control means drives and controls the fuel supply means based on the set fuel supply amount.

<Operation> According to this configuration, the learning correction value area-specific storage unit 1
When the current operating condition corresponds to one of the divided areas, the air-fuel ratio learning correction value stored in the corresponding area of the learning correction value area-specific storage unit is modified and rewritten, and only the one divided area is further modified. A plurality of divided sub-regions are set, and the air-fuel ratio learning correction value is learned as a sub-division learning correction value for each sub-region. That is, even in a state corresponding to one section area of the learning correction value area-based storage means,
When the operating condition changes in the one divided area, the air-fuel ratio learning correction value at that time is stored for each of the more divided areas, and learning is performed for each of the further divided areas of the one divided area. .

Here, when a corresponding region in the divided operation region in the learning correction value region storage unit changes, learning is performed based on an average value of the plurality of subdivision learning correction values stored in the subdivision region storage unit. Since the air-fuel ratio learning correction value of the corresponding divided operation region in the correction value region storage means is rewritten and corrected, learning corresponding to the average correction request in the corresponding region in the learning correction value region storage means is performed. You.

Further, after the rewriting and correction based on the average value, the storage content of the storage unit for each sub-region becomes unnecessary, so that the storage unit for each sub-region further divides only one section area of the storage unit for each learning correction value into a plurality. It is sufficient if there is a capacity for storing the air-fuel ratio learning correction value corresponding to the number of sections to be performed, and an increase in the storage capacity can be suppressed.

<Examples> Examples 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 via an intake duct 3, a throttle valve 4 and an intake manifold 5. A fuel injection valve 6 as a fuel supply means is provided for each cylinder in a branch portion of the intake manifold 5. This fuel injection valve 6
Is a normally-closed electromagnetic fuel injection valve that is energized by a solenoid to open the valve, and is de-energized and closed by being energized by a drive pulse signal from a control unit 12, which will be described later. Injects and supplies fuel that is pressure-fed from a non-fuel pump and adjusted to a predetermined pressure by a pressure regulator.

An ignition plug 7 is provided in each combustion chamber of the engine 1 to ignite and burn an air-fuel 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 three-way catalyst 10 oxidizes and reduces NOx, CO, and HC to convert them into harmless components. Since it is necessary to control the fuel ratio to the stoichiometric air-fuel ratio, in the air-fuel ratio feedback control described later, control is performed such that the stoichiometric air-fuel ratio is the target air-fuel ratio.

The control unit 12 includes a microcomputer including a CPU, a ROM, a RAM, an A / D converter, and an input / output interface, receives input signals from various sensors, performs arithmetic processing as described below, The operation of the fuel injection valve 6 is controlled.

As the various sensors, an air flow meter 13 is provided in the intake duct 3, and outputs a signal corresponding to the intake air flow rate Q of the engine 1.

Further, when the crank angle sensor 14 is provided and the number of cylinders is four, a reference signal REF for every 180 ° of the crank angle and a unit signal POS for one or two crank angles are output. Here, the engine speed N 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 detecting means is provided at a collecting portion of the exhaust manifold 8, and detects an air-fuel ratio of the intake air-fuel mixture through an oxygen concentration in the exhaust gas. The oxygen sensor
Reference numeral 16 is a known device that detects rich / lean of the actual air-fuel ratio with respect to the stoichiometric air-fuel by utilizing the sudden change in the oxygen concentration in the exhaust gas at 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. 3 to 8, respectively, and performs the air-fuel ratio feedback correction control and the air-fuel ratio for each operation region. Fuel injection amount while executing learning correction control
Ti is set to control the fuel supply to the engine 1.

In this embodiment, the functions of the basic fuel supply amount setting means, the air-fuel ratio feedback correction value setting means, the fuel supply amount setting means, the fuel supply control means, the learning correction means, and the correction means based on the average value are the same as those of the third embodiment. As shown in the flowcharts of FIG. 8 to FIG. 8, the control unit 12 is provided as software. Further, the storage means for each learning correction value area and the storage means for each subdivision area correspond to a RAM backed up by a microcomputer built in the control unit 12 in this embodiment. Further, the operating condition detecting means in the present embodiment corresponds to the air flow meter 13, the crank angle sensor 14, and the like.

First, the outline of the air-fuel ratio control in the present embodiment will be described with reference to the flowchart of FIG. 3, and the details of the control will be described later with reference to the flowcharts of FIGS.

In the flowchart of FIG. 3, first, the basic fuel injection amount Tp (← K × Q / N; K is a constant) is corrected so that the actual air-fuel ratio approaches the target air-fuel ratio (the logical air-fuel ratio in this embodiment). -Fuel ratio feedback correction coefficient LMD is set (step A).

Next, the deviation of the above air-fuel ratio feedback correction coefficient LMD from the target convergence value (initial value) is learned, and the corresponding area on the learning map divided into a plurality of sections by the basic fuel injection amount Tp and the engine speed N is learned. The learning correction coefficient KBLRC (air-fuel ratio learning correction value) is corrected and rewritten in a direction to decrease the deviation (step B).

Then, on the learning map in which the learning correction coefficient KBLRC is stored for each driving area as described above, it is determined whether or not the corresponding driving area is the same as the previous time (step C). When the area to be
The learning correction coefficient KBLRC (subdivision learning correction value) is stored for each subdivision region that further divides the relevant region of the learning map into a plurality (step D).

In other words, even if the vehicle is stopped in one section area on the learning map, the operating conditions (basic fuel injection amount Tp
Alternatively, when the engine rotation speed N) changes and corresponding regions in a plurality of sub-regions further dividing the one divided region change, the learning correction coefficient KBLRC is stored for each. Therefore, for example, when one segmented region on the learning map is further segmented into two segments based on the engine speed N, and when the two segments pass through each of the two sub-regions with a change in the engine speed, The learning correction coefficient KBLRC suitable for each of the two sub-regions is stored.

On the other hand, when the corresponding area on the learning map changes, the learning correction coefficient KBLRC (subdivision) stored in each of the subdivision areas obtained by further dividing one area on the learning map corresponding to before the change. Learning correction value),
The learning correction coefficient KBLRC corresponding to the area before the change is updated and set (step E).

That is, when the vehicle is stopped in one driving region on the learning map, the driving region is further divided into a plurality of regions and each of the regions is learned. The map value is updated based on the average value of the learning correction coefficient KBLRC for each of the plurality of sub-regions stored correspondingly.

As a result, an average correction request value in one segmented operation region of the learning map is learned (see FIG. 12), and when the corresponding region on the learning map changes, as described above. After processing based on the average value,
Learning correction coefficient for each sub-region corresponding to the region before change
Since the data stored in the KBLRC is not required, there is no need to store and hold the data of the subdivision area corresponding to each of the divided areas of the learning map. Since it is sufficient to learn the learning correction value for each subdivision area, the memory capacity can be saved.

Here, a detailed embodiment of the air-fuel ratio learning control shown in the flowchart of FIG. 3 will be described with reference to the flowcharts of FIGS.

The program shown in the flow chart of FIG. 4 controls the air-fuel ratio feedback correction coefficient LMD (air-fuel ratio feedback correction value) multiplied by the basic fuel injection amount Tp (← K × Q / N; K is a constant) by proportional / integral control. The program is established for each revolution (1 rev) of the engine 1. Note that the initial value of the air-fuel ratio feedback correction coefficient LMD is 1, which is a correction term that is multiplied by the basic fuel injection amount Tp, so that the actual air-fuel ratio approaches the stoichiometric air-fuel ratio that is the target air-fuel ratio. The actual air-fuel ratio is changed around the target air-fuel ratio by increasing / decreasing the basic fuel injection amount Tp.

First, in step 1 (referred to as S1 in the figure, the same applies hereinafter), a voltage signal output from the oxygen sensor (O ₂ / S) 16 according to the oxygen concentration in the exhaust gas is read.

Then, in the next step 2, the voltage signal from the oxygen sensor 16 read in step 1 is compared with a slice level (for example, 500 mV) corresponding to the target air-fuel ratio (the stoichiometric air-fuel ratio), and the air of the engine intake air-fuel mixture is compared. It is determined whether the fuel ratio is rich or lean with respect to the target air-fuel ratio.

When it is determined that the voltage signal from the oxygen sensor 16 is larger than the slice level and the air-fuel ratio is rich, the process proceeds to step 3, where it is determined whether the current rich determination is the first time.

When the rich determination is performed for the first time, the process proceeds to step 4 and the air-fuel feedback correction coefficient LMD set up to the previous time is set to the maximum value a. The fact that the rich determination is the first time means that the lean determination has been performed up to the previous time, and the increase control of the air-fuel ratio feedback correction coefficient LMD (= increase correction of the fuel injection amount Ti) has been performed by this. When the rich determination is made, the correction coefficient LMD is controlled to be reduced in the future. Therefore, the value before the reduction control for the first time of the rich determination is the maximum value of the correction coefficient LMD.

In the next step 5, reduction control of the correction coefficient LMD is performed by subtracting a predetermined proportionality constant P from the correction coefficient LMD up to the previous time, so that the air-fuel ratio becomes lean and approaches the target air-fuel ratio. In the next step 6, "1" is set to "addition of P" which is a flag indicating that the proportional control has been executed.

On the other hand, if it is determined in step 3 that the rich determination is not the first time, the process proceeds to step 7, and the value obtained by multiplying the integration constant I by the latest fuel injection amount Ti is used as the correction coefficient LM up to the previous time.
The correction coefficient LMD is updated by subtracting from D and the reduction control of I × Ti is repeated in this step 7 every time this program is executed until the rich state of the air-fuel ratio is eliminated and the air-fuel ratio is reversed to lean. .

When it is determined in step 2 that the actual air-fuel ratio is lean with respect to the target air-fuel ratio (the stoichiometric air-fuel ratio),
As in the case of the rich determination, first, in step 8, it is determined whether or not the current lean determination is the first time, and if it is the first time, the process proceeds to step 9 and the correction coefficient LM up to the previous time is determined.
D, that is, the correction coefficient LMD that has been gradually reduced at the time of rich determination is set to the minimum value b, and in the next step 10, the fuel coefficient is updated by adding the proportionality constant P to the previous correction coefficient LMD. The injection amount Ti is increased, and at step 11, 1 is set to the "addition of P" which is a flag indicating that the proportional control has been executed in the same manner as at step 6.

On the other hand, when it is determined in step 8 that the lean determination is not the first time, the process proceeds to step 12, and the value obtained by multiplying the integration constant I by the latest fuel injection amount Ti is used as the correction coefficient LMD up to the previous time.
To gradually increase the correction coefficient LMD.

When the proportional control of the correction coefficient LMD for the first time of the rich / lean determination is performed, various processes related to the air-fuel ratio learning control after step 13 are further performed.

In this embodiment, as shown in FIG. 9, the entire operation region is divided into a plurality of parameters as the basic fuel injection amount Tp, the engine speed N, and the parameter, and the air-fuel ratio learning correction value is rewritten for each of the divided regions. There are two learning maps that can be stored and retained as much as possible. One learning map divides the entire operation area into 16 unit areas with a 4 × 4 grid, and the other learning map has a 16 × 16 learning area.
The entire operation area is divided into 256 unit operation areas on the grid, and one sectioned operation area of the 4 × 4 grid map is
A 16 × 16 learning map is used to further subdivide into 4 × 4 = 16 areas. Further, the air-fuel ratio learning correction value adapted to the entire operating region that does not divide the operating region as described above is also separately learned and set.

In step 13, it is determined whether or not the count value cnt for determining whether or not the vehicle is stably stopped in one of the 16 operation regions divided by the 4 × 4 grid is zero.

In the flow chart of FIG. 5 described later, when the corresponding operation area in the operation area divided by the 4 × 4 grid changes every predetermined minute time, the count value cnt
Is set to a predetermined value (for example, 4). If it is determined in step 13 that the count value cnt is not zero, the process proceeds to step 14 in which the count value cnt is reduced by one. After stopping in one operation area in the 4 × 4 grid, the count value cnt is corrected by the correction coefficient L.
The count value is decremented by one for each proportional control of the MD. When the count value cnt is zero, it can be regarded that the state is stably stopped in one area of the 4 × 4 grid. It is.

Incidentally, by determining whether or not the count value cnt is zero, it is determined whether or not to perform learning update as described later. is there.

In step 15, it is determined whether or not the air-fuel ratio learning correction value corresponding to each of the divided operation regions of the 4 × 4 grid learning map has been almost learned.

Here, when it is determined that most of the areas have been learned on the learning map of the 4 × 4 grid, it is estimated that the base air-fuel ratio is basically learned and controlled to the target air-fuel ratio in the entire operation area. In this case, proceed to the next step 16, where 16 ×
Based on the flag fz, it is determined whether the corresponding driving area on the learning map of 16 grids is the same as the previous one or has changed.

The flag fz is set in a flowchart of FIG. 5 described later. When 1 is set in the flag fz, it indicates that the corresponding driving area has changed on the learning map of the 16 × 16 grid. , Zero is set, indicating that the corresponding area is the same as the previous time.

Here, when the flag fz is set to zero, the present program is terminated as it is, but when the flag fz is set to 1, the process proceeds to step 17.

In step 17, the difference between the absolute value of the deviation of the latest correction coefficient LMD average value from the target convergence value (= 1.0) and the previous absolute value of the deviation is obtained, and based on the value, the degree of inappropriateness of the learning value is determined. The Δ stress is set to increase as the average value variation of the correction coefficient LMD increases, with reference to a Δ stress map indicating

That is, since the learning of the learning map of the 4 × 4 grid is substantially completed, even if the operating conditions change, if the learning has progressed with high accuracy, the required correction amount due to the change in the operating conditions can be reduced. Corresponding to the difference, the change in the base air-fuel ratio should be suppressed to a small level. Here, if the change in the base air-fuel ratio due to the change in the operating conditions is large and the air-fuel ratio feedback correction coefficient LMD greatly changes from the target convergence value to compensate for this, it is indirectly determined that the learning value is inappropriate. Therefore, Δ stress indicating the degree of inappropriateness of the learning value is set according to the fluctuation level of the correction coefficient LMD.

Further, in step 18, the Δ stress is integrated, and the result of the integration is stored as stress. As will be described later, when the stress exceeds a predetermined value, it is determined that the already learned air-fuel ratio learning correction coefficient is inappropriate, and the learning is restarted from the beginning.

In the next step 19, the flag fz is reset to zero so that 1 is set again when the corresponding area on the learning map of the 16 × 16 grid changes.

Then, in step 20, as shown in FIG. 10, the empirical state of the subdivided region obtained by further dividing the corresponding divided operation region on the learning map of the 16 × 16 grid into two based on the engine rotation speed N is shown. The area counter Acct shown is determined.

That is, in a state where the vehicle is stopped in one segmented operation region on the learning map of the 16 × 16 grid, as described later, only the low rotation side of the subdivided region obtained by further dividing the segmented operation region into two is experienced. In this case, 1 is set in the area counter Acct. When only the high-speed side is experienced, 2 is set in the area counter Acnt. When both areas are experienced, 3 is set in the area counter Acct. (See the flowchart of FIG. 5).

Here, it is determined that 1 is set in the area counter Acct, and only one low-speed side in the divided area is stopped in the one divided operation area on the learning map of the 16 × 16 grid. When experienced, the learning correction coefficient KBLRC2 stored corresponding to the subdivision region on the low rotation side
re1 is set to reg in step 21. Similarly, when the area counter Acct is set to 2 and only the high-rotation-side subdivision area is experienced, re as the learning correction coefficient KBLRC2 stored corresponding to the high-rotation-side subdivision area is performed.
2 is set to reg in step 22. Furthermore, area counter
If Acnt is set to 3 and both the low rotation side and the high rotation side have been experienced, the average value of re1 and re2 stored for each of the subdivision areas is calculated in step 23 as r
Set to eg.

Then, in the next step 24, the learning correction coefficient KBLRC2 of the region [s, t] which has been hitherto applied in the learning map of the 16 × 16 grid is used as the learning correction coefficient KBLRC2 set in the reg.
To update the map value.

That is, the learning correction coefficient KBLRC2 is rewritably stored in each operation region of the learning map of the 16 × 16 grid, and the learning correction coefficient KBLRC2 of the corresponding operation region is determined by the air-fuel ratio feedback as described later. Correction coefficient LMD
Is learned and updated based on the deviation from the target convergence value. Apart from such rewriting and modification of the learning map, two subdivided areas are set by further dividing the relevant operating area into two based on the engine speed N. And the learning correction coefficient KBLR that has been learned and updated when it is included in each of the sub-regions.
C2 is sequentially updated and stored in correspondence with each of the sub-regions.

If the corresponding area changes on the learning map of the 16 × 16 grid, the learning correction coefficient KBLRC2 data for each sub-area corresponding to the area corresponding to the previous time is used.
This is for rewriting the learning map of the 16 × 16 grid.

Here, if either the low rotation side or the high rotation side is experienced in the corresponding region of the learning map of the 16 × 16 grid, the learning is stored corresponding to the corresponding region of the learning map of the 16 × 16 grid. The learning correction coefficient KBLRC2 is the same as the data re1 or re2 corresponding to the subdivision area, and the update setting in step 24 does not substantially take effect, but both the low rotation side and the high rotation side are experienced. In this case, in the corresponding area of the learning map of the 16 × 16 grid, the average correction request values on the low rotation side and the high rotation side are set as the learning correction coefficient KBLRC2.

For example, if you enter a learning area on the 16 × 16 grid learning map from the low rotation side, move to the high rotation side, and then move to another area, the difference in rotation speed even in the same learning area If there is a difference in the correction request, the data of re1 and re2 will be different data corresponding to the difference of the correction request, and if these average values are used as update data in the learning map, the data of the learning map of the 16 × 16 grid The learning corresponding to the average correction request on the low rotation side and the high rotation side in one divided area is performed.

When the above processing is not performed, for example, when the rotation speed is slowly increased, learning of the corresponding area on the learning map of the 16 × 16 grid is learning corresponding to the high rotation side immediately before passing through the area, and this learning is performed. Next, when the vehicle enters the region where such learning is performed from the low rotation speed side, the correction value corresponding to the request for the high rotation speed is learned. A step will be generated.

In order to solve such a problem, a learning map may be provided which divides the driving region into smaller areas than the learning map of the 16 × 16 grid, but in this case, the memory capacity for the learning map is significantly increased. Resulting in. On the contrary,
In the case of the present embodiment, in addition to the memory capacity of the learning map of the 16 × 16 grid (256 learning correction coefficients KBLRC2), each of the subdivided areas that subdivides only one corresponding area of the 16 × 16 grid map into two If there is a memory capacity capable of storing two learning correction coefficients KBLRC2 corresponding to the above, an air-fuel ratio controllability substantially equivalent to the case where learning is performed by dividing into 16 × 16 × 2 = 512 areas can be obtained. , And can cope with a correction request for each finer area.

Further, even if the user does not experience all of the 512 regions, the learning of the 256 learning regions of the 16 × 16 grid proceeds, so that the learning progress is not hindered in order to respond to the fine correction request for each driving region.

In the next step 25, the area counter Acct is reset to zero, and in the next step 26, which of the sub-areas that divides the relevant area on the learning map of the 16 × 16 grid into two is determined. The indicated area number Ano is reset to zero to prepare for learning in a newly applicable area in the learning map of the 16 × 16 grid.

The program shown in the flowchart of FIG. 5 is an air-fuel ratio learning program for each operation region, and is executed every predetermined short time (for example, 10 ms).

In step 31, it is determined whether or not the addition of P is a flag which is set to 1 when the proportional control of the air-fuel ratio feedback correction coefficient LMD is performed in the flowchart of FIG. Go to step 32 P
After the minute addition is reset to zero, various processes are performed by the present program, and when it is zero, the present program is terminated as it is. That is, the program shown in the flowchart of FIG. 5 is executed once every time the air-fuel ratio feedback correction coefficient LMD is proportionally controlled.

In step 32, the addition of the P component is reset to zero, and in the next step 33, a flag indicating whether or not a learning correction coefficient KBLRCφ (initial value 1), which is an air-fuel ratio learning correction value common to all operation regions, has been learned. Fφ is determined.

Here, when the flag Fφ is zero and the learning correction coefficient KBLR
If the learning of Cφ has not been completed, the routine proceeds to step 34, where the average value of the maximum and minimum values a and b of the correction coefficient LMD (←
It is determined whether (a + b) / 2) is approximately 1.

If (a + b) / 2 is not substantially 1, the routine proceeds to step 36, where the target convergence value Target of the correction coefficient LMD is calculated from (a + b) / 2.
A value obtained by multiplying a value obtained by subtracting the initial value (1.0 in this embodiment) by a predetermined coefficient X is added to the previous learning correction coefficient KBLRCφ, and the result of the addition is set as a new learning correction coefficient KBLRCφ.

In step 36, a 4 × 4 grid map and a 16 × 16 grid map are used.
The learning correction coefficient KBLRC1 and the learning correction coefficient KBLRC2 stored for each of the divided operation areas of the grid map are all set to 1.0, which is the initial value, to return to the initial state. Therefore, when a new learning update of the learning correction coefficient KBLRCφ is performed, a 4 × 4 grid map and a 16 × 16
Even if the learning value is stored in the grid map, the learning is performed from the learning correction coefficient KBLRCφ after resetting the data.

On the other hand, if it is determined in step 34 that (a + b) / 2 is substantially 1, the flag Fφ is set to 1 in step 35 to complete learning of the learning correction coefficient KBLRCφ corresponding to the entire operation range. That is, in other words, the learning correction coefficient KBLR
It is possible to determine that the air-fuel ratio feedback correction coefficient LMD has converged to approximately 1 as a result of learning and updating Cφ.

If it is determined in step 33 that the flag Fφ is 1, the learning correction coefficient KB corresponding to the entire operation range is set.
This indicates that the learning of LRCφ has been completed, so that the air-fuel ratio learning is performed for each of the divided operating regions based on the basic fuel injection amount Tp and the engine speed N.

First, on the learning map of the 16 × 16 grid, an area to which the current driving condition is applicable is specified.

In step 37, a counter value i for determining the order of the 16 grids on the 16 × 16 grid learning map is set to zero, and the next step
At 38, it is determined whether or not the counter value i exceeds 15, and if not, at step 39, the threshold value Tp [i] of the basic fuel injection amount Tp corresponding to the count value i is updated. The calculated basic fuel injection amount Tp is compared.

When it is determined in step 39 that the latest basic fuel injection amount Tp is smaller than the threshold value Tp [i], the count value i at that time is set in the area where the latest basic fuel injection amount Tp corresponds on the 16 grid. The process proceeds to step 42 and subsequent steps assuming that the position is indicated. That is, the maximum basic fuel injection amount Tp of each segmented region is set in advance as a threshold value Tp [i], and the latest basic fuel injection amount Tp and the threshold value Tp [i] are set as threshold values. The value i at the time when Tp [i]> Tp has been satisfied for the first time after the comparison is performed in ascending order is set as the value indicating the position of the Tp region to which the current Tp corresponds.

If it is determined in step 39 that Tp [i] ≦ Tp, the process proceeds to step 40, where the count value i is increased by one, and Tp [i], which is one step larger, is compared with the latest Tp. To do.

When the count value i is counted up to 16 in step 40, the basic fuel injection amount larger than the initially set maximum value of the basic fuel injection amount Tp range divided into 16 grids (16 areas) from 0 to 15 In this state, Tp is calculated. At this time, i is set to 15 in step 41, and then the process proceeds to step 42, where the maximum Tp of the initially set Tp region is set.
Assume that the region contains the current Tp.

When the region including the current Tp is specified as described above and the process proceeds to step 42, I
And the latest area i.

Here, when it is determined that 1 ≠ i, it is when the basic fuel injection amount Tp changes and the corresponding area changes,
At this time, in step 43, the flag fz is set to 1 so that the change of the corresponding area on the 16 × 16 grid map is determined. In the next step 44, I indicating the relevant area before the change is set to t. So that even if the area position I is updated to the data indicating the latest applicable area, the applicable area before the change can be determined by the aforementioned t.

Further, in step 45, when a certain area of the 16 × 16 grid map is further divided into two sub-areas, an area number indicating which of the sub-areas is included in the sub-area is reset to zero, and In step 46, the area counter Acct is reset to zero so that learning in the corresponding area after the change is newly performed.

In step 47, the corresponding region position i of Tp newly obtained this time is finally set to I.

Next, for 16 blocks divided by engine speed N,
In the same manner as the block determination of the basic fuel injection amount Tp, the block position including the latest engine speed N is determined by the count value k.

First, in step 48, the count value k is initially set to zero. Until it is determined in step 49 that the count value k has exceeded 15, the threshold value N in step 50 has been reached.
Comparison with [k] is performed. If N [k] ≦ N, the count value k is increased by 1 in step 51, and a count value k satisfying N [k]> N is obtained for the first time. When the count value k exceeds 15, the process proceeds to step 52, where the maximum value 15 is set as the count value k, and the process proceeds to step 53.

When the region position to which the latest engine rotational speed N corresponds is obtained as k, in step 53, it is determined whether or not the region has changed in the same manner as in step 42, by comparing the previous region with the corresponding region position K and the latest corresponding region. When the corresponding area changes, the flag fz is set to 1 in step 54, and the area position K up to the previous time is set to s in step 55. The previous corresponding area can be determined.

In step 56, the latest engine rotational speed N is included on the low rotational speed side in the region at one section quantity on the 16 × 16 grid map determined to include the latest engine rotational speed N. Is included in the high rotation side, in other words, one sectioned area of the 16 × 16 grid learning map is further divided by 2
It is determined which of the sub-regions to divide into.
This determination is made based on the rotation speed range N [k−
1] to N [k] center value N [k] -N [k-1] / 2 + N
This is performed by comparing the magnitude of [k-1].

Here, when it is determined that the area is included on the low rotation side of the area, it is determined in step 57 whether the area counter Acct is 2 or 3.

If the area counter Acct is not 2 or 3, the process proceeds to step 58, and 1 indicating that the low rotation side of the area has been experienced is set to Acct. As will be described later, when the high speed side of the corresponding area is experienced, 2 is set to the above-described Acct. Therefore, when the operation shifts from the high speed side to the low speed side in the corresponding area, the Acct is set to 2 in the step 57. In this case, the process proceeds to step 59 and the Ac
By setting 3 to nt, it is determined that both the low rotation speed side and the high rotation speed side have been experienced due to the fluctuation of the rotation speed in the corresponding region.

If it is determined in step 57 that Acct is set to 3, it indicates that the relevant area has already experienced the subdivision areas on both the low rotation side and the high rotation side.
Set 3 as it is in nt.

Then, in step 60, the current operating condition is 16 × 16
In order to indicate that it is included in the subdivision region on the low rotation side in the corresponding region on the grid map, 1 is added to the region number Ano.
Is set.

On the other hand, if it is determined in step 56 that the engine rotation speed N is included on the high rotation side in the corresponding area, it is determined in step 61 whether the area counter Acct is 1 or 3 and 1 or 3 is determined.
If none of the above is satisfied (when the vehicle enters the corresponding area from the high rotation side), two are set in Actnt in step 62. Also,
If it is determined in step 61 that Acct is set to 1 or 3, the step is performed to indicate that the user is experiencing both the low rotation side and the high rotation side subdivision regions of the region.
At 63, 3 is set to the above-mentioned Acct.

Then, in step 64, the current operating condition is 16 × 16
In order to indicate that it is included in the sub-region on the high rotation side in the corresponding region on the grid map, 2 is added to the region number Ano.
Is set.

After setting the area number Ano in step 60 or step 64, k obtained as the area position including the latest engine speed N in step 65 is finally set to K. In this manner, which operating region of the learning map divided into 256 operating regions of 16 × 16 using the basic fuel injection amount Tp and the engine speed N as parameters is included in the Tp region. The coordinates are represented by coordinates [K, I] represented by the position I and the region position K at the position N.

If the corresponding position of the 16 × 16 grid is determined as described above, as shown in FIG. 9, one operating area in the 4 × 4 grid operating area map is 4 × in the 16 × 16 grid operating area. Since 4 = 16 areas are divided as one block, a position to which the current driving condition corresponds in the learning map of the 4 × 4 grid can be specified based on the I and K.

That is, in step 66, the block number I of the Tp is set to 4
, And set the integer value obtained by rounding down the decimal point of the result to A. In step 67, similarly, divide the block number K of N by 4 to round down the decimal point of the result. Is set to B. This allows
The area position on the 4 × 4 grid map to which the current operating condition corresponds is represented by the coordinates [B, A].

In the next step 68, using [B, A] indicating the corresponding area position on the 4 × 4 grid map, in order to determine whether or not the corresponding driving area has changed on the 4 × 4 grid, A
Is multiplied by 16 and B is added, and the result is set in AB.

Then, in step 69, the calculated AB, AB
By comparing _OLD with the latest AB, it is determined whether or not the area corresponding to the current time on the 4 × 4 grid map is the same as the previous time. If AB ≠ AB _OLD and the corresponding area on the 4 × 4 grid map is different from the previous area, step 70
Sets a predetermined value (for example, 4) to the count value cnt.

In step 71, for the next determination in step 69, AB calculated this time in step 68 is set as the previous value and AB
Set to _CLD .

In step 72, in the 4 × 4 grid map, [B, A]
F indicating whether or not the air-fuel ratio learning has been completed in the operation region including the current operation condition indicated by the coordinates
[B, A] is discriminated, and if this flag F [B, A] is zero and learning is not completed in one driving area of the 4 × 4 grid map including the current driving condition, step 73 is executed. Proceed to.

In step 73, it is determined whether or not the count value cnt is zero. If the count value is not zero and the corresponding area in the 4 × 4 grid map fluctuates, the program is terminated as it is, and the count value cnt is zero. The process proceeds to step 74 only when the vehicle is stably stopped in the corresponding operation area on the 4 × 4 grid map.

In step 74, the average value (a + b) / 2 of the maximum and minimum values a and b of the air-fuel ratio feedback correction coefficient LMD sampled in the flowchart of FIG. 4, that is, the correction coefficient L
The progress of learning is determined based on whether or not the center value of the MD is near the initial value (= 1) which is the target convergence value of the correction coefficient LMD. Proceed to step 76.

In step 76, the learning correction coefficient KBL stored in the 4 × 4 grid map corresponding to the current [B, A] area
The target convergence value is calculated from the average of the maximum and minimum values a and b for RC1.
The value obtained by subtracting Target (1.0 in this embodiment) is given by a predetermined coefficient X1.
Are added, and the result is added to the learning correction coefficient KBLRC1 corresponding to the current operation area [B, A] on the 4 × 4 grid map.
As a new update setting.

If (a + b) / 2 is determined to be approximately 1 in step 74, it means that learning in the operation area of the 4 × 4 grid map including the current operation conditions has been completed, and thus step 75
Then, the flag F [B, A] is set to 1 so that it is determined that the learning is completed for the area where the flag F [B, A] is 1.

During the learning in the area [B, A] on the 4 × 4 grid map, the learning correction in each of the 4 × 4 = 16 areas included in the [B, A] in the finer 16 × 16 grid map is performed. All coefficients KBLRC2 are reset to an initial value of 1.0 in step 77.

As described above, when there is a region in which the learning is not completed in the 4 × 4 grid map, the predetermined ratio of the deviation from the target value (Traget) of (a + b) / 2 when the operation region is stable is
By adding and updating the learning correction coefficient KBLRC1 stored up to that point, the air-fuel ratio feedback correction coefficient
The target air-fuel ratio can be obtained by correction using the learning correction coefficient KBLRC1 instead of LMD, and learning of the operating region is completed when the air-fuel ratio feedback correction coefficient LMD substantially converges to the initial value 1, which is the target convergence value. And

On the other hand, in step 72, when the flag F [B, A] is determined to be 1, and the learned correction coefficient KBLRC1 that has been learned is stored in the currently applicable operation area of the 4 × 4 grid map, the learning correction coefficient This time 4 × 4 where KBLRC1 is stored
The operation area [B, A] on the grid map is further subdivided into 16 areas, and the operation shifts to learning in the corresponding area of the 16 × 16 grid learning map.

In step 78, the average value of the correction coefficient LMD (a +
It is determined whether or not (b) / 2 is substantially equal to 1 of the target convergence value. In the unlearned state where (a + b) / 2 is not approximately 1 but needs to be corrected by the air-fuel ratio feedback correction coefficient LMD. If so, go to step 79.

In step 79, the target convergence value Target is calculated from (a, b) / 2.
The value obtained by multiplying the value obtained by subtracting (1.0 in the present embodiment) by the predetermined coefficient X2 is stored up to that time in the 16 × 16 grid learning map corresponding to the driving area [K, I] in which the driving condition is included. Is added to the previously set learning correction coefficient KBLRC2 [K, I], and the result of the addition is added to a new correction coefficient KBLRC in the operation region [K, I].
Update setting as 2 [K, I].

Then, in the next step 80, the area number Ano
To determine which of the two sub-regions the corresponding region [K, I] is subdivided into based on the rotation speed N. If the region number Ano is 1 and the corresponding region When [K, I] is on the low rotation side, in step 81, r
In e1, the learning correction coefficient KBLRC2 newly obtained in step 79
When the area number Ano is 2 and is on the high rotation side of the corresponding area [K, I], the learning correction coefficient KBLRC2 newly obtained in step 79 is set in re2 in step 82.

That is, the learning correction coefficient KBLRC corresponding to the corresponding area [K, I]
When update setting 2 is performed, the learning correction coefficient KBLRC2 is separately stored depending on whether the rotation speed is on the low rotation side or the high rotation side in the corresponding region (K, I), and the corresponding region (K, I) is stored.
When the rotation speed fluctuates in both the low rotation side and the high rotation side in (I), the learning correction coefficient KBLRC2 suitable for the low rotation side of the corresponding region (K, I) is stored as re1. The learning correction coefficient KBLRC2 suitable for the high rotation speed side is stored as re2.

Then, as described above, when the corresponding region in the learning map of the 16 × 16 grid changes, the learning correction coefficient KBLRC2 of the learning region corresponding to before the change is rewritten based on the storage data re1 and re2 for each of the sub-regions. The map learning is performed as an average level of the correction request on the low rotation side and the high rotation side in one section area of the learning map of the 16 × 16 grid that has been corrected.

On the other hand, when it is determined in step 78 that the average value (a + b) / 2 of the correction coefficient LMD substantially matches the target convergence value Target, the routine proceeds to step 83, where the current value of the 16 × 16 grid map is determined. Flag FF (K, I) so that it is determined that the learning of the operation region [K, I] including the operation condition of
I] is set to 1.

Then, after step 84, it is determined that the learning has been completed this time, and 1 is set to the corresponding flag FF [K, I].
Based on the predetermined operation area [K, I] on the 6 × 16 grid map, if there is an operation area in which learning has not been completed in an operation area adjacent to this area [K, I] (see FIG. 11), Control is performed to update and store the same value as the learning correction coefficient KBLRC2 stored in the operation region corresponding to the current operation region [K, I].

In step 84, values obtained by subtracting I from [K, I] indicating the area position including the current driving condition in the 16 × 16 grid map are set to m and n, and in the next step 85, m = K + 2. It is determined whether or not there is.

When the process proceeds from step 84 to step 85, the determination of NO is made in step 85. Therefore, the process proceeds to step 86 and the learning of one operating region of the 16 × 16 grid indicated by [m, n] is completed. Flag FF [m, n] is 1
Or zero.

Here, when the flag FF [m, n] is zero and learning is not completed, the process proceeds to step 87. This step
At 87, the area address [m, n] on the 16 × 16 grid is converted to the area address [m / 4, n / 4] on the 4 × 4 grid, and this is determined to be the current 4 × 4 area. It is determined whether or not it matches the area [B, A] on the grid map.

That is, [K, I] is an area included in [B, A],
The area around [K, I] is [B, A] on a 4 × 4 grid map.
May be included in another area adjacent to
If the area is included in the same [B, A] ([m / 4, n /
4] = [B, A]), the process proceeds to step 88, where the learning correction coefficient KBLRC corresponding to the [K, I] region determined to have been learned this time.
2 is used as a learning value in the [m, n] region as it is.

On the other hand, in step 87, [m, n] adjacent to [K, I]
When it is determined that the area is included in a different area on the × 4 grid map, the process proceeds to step 89, and the learning correction coefficient KBLRC2 calculated by the following equation is stored in the area [m, n].

KBLRC2 [m, n] ← KBLRC1 [B, A] + KBLRC2 [K, I]-KBLRC1 [m / 4, n / 4] Since [m, n] is an area adjacent on a 16 × 16 grid,
The final correction request is based on the presumption that they should be close to each other.
Since the learning correction coefficient KBLRC1 of the grid map is different, the sum of different KBLRC1 [B, A] and KBLRC [m / 4, n / 4] is
It is set as an approximation as in the following equation.

KBLRC1 [B, A] + KBLRC2 [K, I] = KBLRC1 [m / 4, n / 4] + KBLRC2 [m, n] When the [m, n] area has been learned as described above, this learning value Without updating, and when it is not yet learned, if KBLRC2 [m, n] is updated and set based on KBLRC2 [K, I], in step 90, m is increased by 1 and the process returns to step 85 again. Until m = K + 2, that is, while n is constant, m is moved within a range of ± 1 around K, and learning is performed and unlearned is determined for each operation region.

Then, if it is determined in step 85 that m = K + 2 as a result of the 1-up processing of m in step 90, the process proceeds to step 91, where it is determined whether or not n = I + 2.
If ≠ I + 2, m is again set to K−1 in step 92.
After increasing n by 1 in the next step 93,
Proceed to step 86.

Therefore, first, n = I-1 and m is moved within a range of ± 1 around K to determine the adjacent area. Next, n = I and m around K. Move in the range of ± 1, and then move m in the range of ± 1 centered on K with n = I + 1, resulting in 8 surrounding [K, I].
When learning has not been performed for one operation region (see FIG. 11), a value based on the learning correction coefficient KBLRC2 [K, I] is stored as the learning correction coefficient KBLRC2 [m, n] for that operation region.

If the learning result of the learned area is applied to the surrounding unlearned area in this way, the driving area is subdivided like a learning map of a 16 × 16 grid. Even when the amount of air flow is small, it is possible to prevent a large step from occurring in the air-fuel ratio controllability between the operation regions.

When it is determined in step 91 that n = I + 2, it means that the determination processing for all eight operating regions surrounding [K, I] has been completed. At this time, the process proceeds to step 79, where the current region [ [K, I], the learning correction coefficient KBLRC2, which has already been determined to have been learned, is updated, thereby further improving the learning accuracy.

As described above, in the present embodiment, first, the learning correction coefficient KBURCφ corresponding to the entire operation region is learned, and then learning is performed for each operation region divided by the learning map of the 4 × 4 grid.
Further, as for the area which has been learned by the learning map of the 4 × 4 grid, the area is further divided into 4 × 4 grids, and learning (learning of the 16 × 16 grid learning map) is performed, so that it is large. The operation proceeds from the operating region to the learning in the small operating region.Learning in the large operating region ensures the convergence of the air-fuel ratio. Therefore, it is possible to accurately cope with the difference in the required correction value due to the difference in the operation region.

Needless to say, as described above, one learning region in the learning map of the 16 × 16 grid is further divided into two, and a learning correction value suitable for each sub-region is stored, and when exiting from the one learning region, Since the learning correction value of the learning region is updated based on the average value of the learning correction values for each of the sub-regions, the learning map is further improved than the learning map of the 16 × 16 grid without significantly increasing the memory capacity. It is possible to perform learning substantially corresponding to a correction request for each fine operation region.

Three learning correction coefficients KBLR learned as described above
The final setting of the air-fuel ratio learning correction coefficient KBLRC based on Cφ, KBLRC1, KBLRC2 is performed according to the flowchart of FIG.

The flowchart in FIG. 6 is processed as a background job. First, in step 101, the learning correction coefficient KBLRC1 stored in the area [B, A] of the 4 × 4 grid map is read, and the next step is executed. In step 102, the learning correction coefficient KBLRC2 stored in the area [K, I] of the 16 × 16 grid map is read. Note that B × 4 + A and K × 16 + I shown in the flowchart are for converting addresses indicating respective grid positions into addresses on the memory.

In step 103, KBLRCφ + KBLRC1 + KBLRC2−2.0 → K
The final learning positive coefficient KBLRC is set as BLRC.

The learning correction coefficient KBLRC set in the flowchart of FIG. 6 is used for calculating and setting the fuel injection amount Ti in the program shown in the flowchart of FIG.

The program shown in the flowchart of FIG. 7 is executed every predetermined minute time (for example, 10 ms). First, in step 111, the intake air flow rate Q detected by the air flow The engine speed N calculated based on the detection signal is input.

Then, in the next step 112, the basic fuel injection amount Tp corresponding to the intake air flow rate Q per unit rotation based on the intake air flow rate Q and the engine speed N inputted in step 111.
(← K × Q / N; K is a constant).

In the next step 113, the basic fuel injection amount Tp calculated in step 112 is subjected to various corrections to calculate the final fuel injection amount (fuel supply amount) Ti. Here, the correction value used to correct the basic fuel injection amount Tp is learned and set for each region according to the flowchart of FIG. 5, and is calculated according to the learning correction coefficient KBLRC finally set in the flowchart of FIG. 6 and the flowchart of FIG. Air-fuel ratio feedback correction coefficient LMD, cooling water temperature Tw detected by water temperature sensor 15
Various correction coefficients COEF set including a basic correction coefficient based on the above, a post-start increase correction coefficient, and the like, a correction amount Ts for correcting a change in the effective injection time of the fuel injection valve 6 due to a change in the battery voltage, and Ti The final fuel injection amount Ti is updated every predetermined time as ← Tp × LMD × KBLRC × COEF + Ts.

At a predetermined fuel injection timing, the control unit 12 outputs to the fuel injection valve 6 a drive pulse signal having a pulse width corresponding to the fuel injection amount Ti calculated latest according to the program shown in the flowchart of FIG. , The amount of fuel supplied to the engine 1 is controlled.

The program shown in the flowchart of FIG.
This is a program for performing processing based on “stress” indicating the degree of inappropriateness of the learning correction value sampled according to the flowchart of FIG. 4, and is executed as a background job (BGJ).

In step 121, the stress (air-fuel ratio deviation degree) integrated in step 18 of the flowchart of FIG. 4 is compared with a predetermined value (for example, 0.8), and the air-fuel ratio deviation degree when learning is almost completed is completed. Is greater than or equal to a predetermined value.

Here, when the stress exceeds a predetermined value, it is determined that although the learning is almost completed, the learning result is inappropriate and an air-fuel ratio deviation has occurred, and learning (repetition of learning) is performed again. The process proceeds to step 122 in order to

In step 122, flags Fφ, F [0, 0] to Fn for determining whether or not the air-fuel ratio learning of each operation region has been completed.
F [3,3], FF [0,0] to FF [16,16] are all reset to zero so that learning from the learning correction coefficient KBLRCφ corresponding to the entire operation range is performed and stress is reduced to zero. Reset.

In this way, if the degree of the deviation (stress) of the air-fuel ratio feedback correction coefficient LMD from the reference value becomes larger than a predetermined value, the learning is performed again. When the fuel ratio changes abruptly, learning is performed again for each large operation range, so that the air-fuel ratio can be quickly converged.

In this embodiment, the learning map of the 4 × 4 grid and the 16 × 16 learning map are used.
A learning map of the lattice is provided.
A configuration having only a 4-lattice learning map may be used.
In this case, one learning region of the learning map of the 4 × 4 grid is divided into a plurality of regions, and the air-fuel ratio learning correction value suitable for each sub-region is stored as in the present embodiment, and the corresponding region is changed. Then, the air-fuel ratio learning correction value in the learning region before the change may be updated as the average value of the air-fuel ratio learning correction value for each of the sub-regions.

Further, in the present embodiment, one learning region in the learning map is divided into two based on the engine speed N. However, the number of segments is not limited to two. One learning area may be subdivided into four or more based on the injection amount Tp.

Furthermore, in the present embodiment, the basic fuel injection amount Tp and the engine rotation speed N are used as parameters for dividing the operation region in the learning map for each operation region. However, the present invention is not limited to these. Clearly, conditions may be used.

In addition, a learning map may be provided by dividing the driving region using only one as a parameter of the driving condition.
The operation region may be divided by three or more operation condition parameters.

<Effects of the Invention> As described above, according to the present invention, when air-fuel ratio learning is performed for each of a plurality of divided operation regions, only the corresponding region among the plurality of divided operation regions is further divided into a plurality of sub-regions. The air-fuel ratio learning correction value is learned and stored for each sub-region, and when exiting from the learning region, the air-fuel ratio learning correction value corresponding to the learning region is determined based on the average value of the sub-division learning correction values for each sub-region. Since the fuel ratio learning correction value is rewritten and corrected, a large amount of air-fuel ratio learning corresponding to differences in required corrections due to differences in operating conditions in the divided operating regions in learning for each of the divided operating regions is greatly stored. There is an effect that the present invention can be realized without increasing the capacity.

[Brief description of the drawings]

FIG. 1 is a block diagram showing the configuration of the present invention, FIG. 2 is a system schematic diagram showing an embodiment of the present invention, FIGS. 3 to 8 are flowcharts showing control contents in the above embodiment, respectively. FIG. 11 to FIG. 11 are diagrams showing the state of the learning map in the embodiment, respectively, and FIG. 12 compares the state of learning in a situation where the engine rotational speed slowly increases in the case of the conventional learning and the case of the embodiment. FIG. 13 is a time chart showing an example of the air-fuel ratio deviation characteristic. 1 ... engine, 6 ... fuel injection valve, 12 ... control unit, 13 ... air flow meter, 14 ... crank angle sensor, 15 ... water temperature sensor, 16 ... oxygen sensor

Claims (1)

(57) [Claims] An operating condition detecting means for detecting an engine operating condition including at least an operating parameter related to an amount of air taken into an engine, and a basic fuel supply based on the engine operating condition detected by the operating condition detecting means. Basic fuel supply amount setting means for setting the amount, air-fuel ratio detecting means for detecting the air-fuel ratio of the engine intake air-fuel mixture, and comparing the air-fuel ratio detected by the air-fuel ratio detecting means with the target air-fuel ratio to determine the actual air-fuel ratio. Air-fuel ratio feedback correction value setting means for setting an air-fuel ratio feedback correction value for correcting the basic fuel supply amount so as to bring the air-fuel ratio closer to the target air-fuel ratio; and an operating region divided into a plurality of regions based on engine operating conditions. A learning correction value area storing means for separately storing rewritable air-fuel ratio learning correction values for correcting the basic fuel supply amount; Learning correction for learning the deviation of the value from the target convergence value and correcting and rewriting the air-fuel ratio learning correction value of the corresponding divided operation region stored in the learning correction value region-specific storage means in the direction of decreasing the deviation. Means, only the operating region corresponding to the learning correction value area storage means is further divided into a plurality of subdivision areas, and the air-fuel ratio learning correction value corrected by the learning value correction means is divided into the plurality of subdivision areas. Storage means for each sub-area that temporarily stores as a sub-division learning correction value in association with the corresponding sub-area, and a corresponding one of a plurality of sub-operation areas in the learning correction value area storage means. Is changed, an average value of a plurality of subdivision learning correction values for each subdivision region stored in the subdivision region storage unit is obtained, and a corresponding one of the learning correction value region storage units is determined based on the average value. An average value for rewriting and correcting the air-fuel ratio learning correction value in the divided operation region; and the basic fuel supply amount, the air-fuel ratio feedback correction value, and the corresponding classification stored in the learning correction value region-specific storage unit. Fuel supply amount setting means for setting a final fuel supply amount based on the air-fuel ratio learning correction value in the operating region; and driving control of the fuel supply means based on the fuel supply amount set by the fuel supply amount setting means. An air-fuel ratio learning control device for an internal combustion engine, comprising: fuel supply control means.