JPS60111034A - Air-fuel ratio control system - Google Patents

Abstract

PURPOSE:To rationalize supplement and change operation of a coefficient of compensating a range by learning by making up a memory range of compensating coefficients from blocks of data used for control, obtained as a result from learning and obtained as a result from learning at the point just before the latest learning results. CONSTITUTION:After shifting to an engine control using a constant learning map wherein learning coefficient Kl constantly controlling a fuel injection timing Ti is stored, the learning coefficient Kl within the operation range corresponding to the constant learning map and a buffer map is amended with a new coefficient every time when the new learning coefficient Kl is obtained by learning in the specified operation range. Those new data are called D' and C' and every time when the amendment is performed to refill a new coefficient, a control coefficient alpha is temporarily made to be 1.0 and C' which is written in a buffer map is compared with data C which is stored in a comparison map. If the number of the differences between the coefficients in those ranges reaches a specified number (m), the routine is repeated. Air-fuel ratio control wherein the average of the cotrol coefficient alpha is kept 1.0 by the learning coefficient Kl may be performed, thereby providing higher responsibility and restraining exhaust gas from getting noxious.

Description

[Detailed Description of the Invention] [Field of Application of the Invention] The present invention relates to an electronic fuel supply control device for automobile gasoline engines, etc., and in particular has a learning function and always performs control under optimal control parameters. The present invention relates to a control device that can perform the following functions.

[Background of the invention]

In an internal combustion engine such as a gasoline engine (hereinafter simply referred to as the engine), the amount of fuel supplied to intake air is maintained at a predetermined ratio so that the ratio (A/F - air fuel ratio) is always maintained correctly. There is a need.

Therefore, in the past, a predetermined air-fuel ratio was obtained by measuring the intake air flow rate and controlling the amount of fuel supplied accordingly, but this method was insufficient when considering exhaust gas regulations. precise control cannot be obtained.

For this reason, 0. So-called 02 feedback control, which uses a sensor to detect the state of exhaust gas and perform feedback control on the amount of fuel supplied, has come into use.

In other words, in this 02 feedback control, the fuel supply amount determined by the above-mentioned intake air flow rate is used as the basic supply amount, and this basic supply amount is corrected by feedback so that the output air-fuel ratio converges to a predetermined value. As a result, even if the air-fuel ratio cannot be controlled correctly by just controlling the basic supply amount described above due to changes in engine operating conditions, etc., the engine will always maintain a predetermined air-fuel ratio during operation. You can do what you do.

An example of an engine control device equipped with such 02 feedback control is shown in FIG.

In FIG. 1, 1 is an electronic control unit based on a microcomputer system, 2 is an engine, and 3 is attached to the exhaust pipe of the engine, which determines the state of the output air-fuel ratio from the oxygen concentration in the exhaust gas. 4 is an injector for fuel injection attached to the intake pipe of the engine.

The electronic control device 1 obtains information such as the intake airflow of the engine (ftQa), its rotational speed N, the temperature of the cooling water, and the battery voltage from sensors (not shown) to know the operating state of the engine. The injector 4 is corrected by the signal from the injector 4 and injects fuel.

Since the fuel injection by the injector 4 at this time is performed periodically and intermittently in synchronization with the rotation of the engine, the fuel supply amount is controlled by controlling the fuel injection time per injection by the injector 4. If this injection time is TL, this Th is determined by the following equation.

T4=to-T・p・α number ΣKL ・・・・・・・・・・・・
・・・・・・(1) Tp=, -Q point ・・・・・・・・・
......(2) K: Coefficient determined by the injector Tp: Basic fuel injection time α: Air-fuel ratio control coefficient KL: Various correction coefficients Qa: Intake air flow rate N: Engine rotation speed (rotational speed) This (2 ) As is clear from the equation, the basic fuel injection time T
Since p is determined by the operating condition of the engine, this is the basic supply section and is 0. In feedback, the control coefficient α is changed to 0. The output of the sensor 3 is made to repeat rich and lean so that the average output air-fuel ratio becomes a predetermined value, that is, the stoichiometric air-fuel ratio (A, /F=14.7'').

Therefore, if the basic injection time Tp is maintained in an ideal state, the candy with the control coefficient α will be in a state of swinging around 1.0, and the average value will be 1.0. , Basic Q7! If the air-fuel ratio due to injection time 'fp becomes 10% lean, in order to correct this, the value of the control coefficient α is set to 7, 2, 10 so that it swings around 1.1.
When it becomes % rich, it will be able to swing around 0.9, and the output air-fuel ratio will work in the ideal state, and even if the air-fuel ratio given by the basic injection time 1'p deviates from the ideal state, it will always oscillate around 0.9. It is possible to maintain the output air-fuel ratio in an ideal state and prevent exhaust gas from deteriorating.

By the way, even when this Q feedback control is applied, there is a practical limit to its response speed, so
If the air-fuel ratio due to the above-mentioned basic supply amount suddenly changes, the control will not be able to follow the sudden change in the air-fuel ratio, and will deviate from the stoichiometric air-fuel ratio in a transient state until the average value of the output air-fuel ratio converges to a predetermined value. This results in deterioration of exhaust gas. Such a sudden change in the air-fuel ratio due to the basic supply amount can often occur when the engine shifts from a sudden acceleration state to an engine braking state.

Therefore, in order to eliminate problems with the feedback control method in such cases, the operating state of the engine is divided into multiple regions according to the rotation speed and intake air flow rate, and a correction coefficient for the basic supply amount is set in advance for each operating region. By setting the basic supply amount in advance and controlling it by correcting the basic supply amount using the engine operating range and the correction coefficients of K and K, the required O
A control method in which the amount of control caused by t-feedback hardly changes has been proposed and has come into use.

In this method, the injection time 1゛L by the injector 4 is determined by the following equation.

TQ = ni-Tp・α・Kr・2℃・・・・・・・・・
- Song... (3) Kr: Area correction coefficient And, in this method, the engine speed change range and the intake air amount change range are each divided into nine JOI7) ranges, and 100 pieces of each combination are created. , and control coefficient α = 1 for each operating region.
．． 0, that is, the same state as when no feedback is performed, determine the region correction coefficient Kr that will give you the stoichiometric air-fuel ratio (-14, 7), store it in a ROM, etc., and start the engine operation. If you read it out from time to time and use it to calculate the injection time TL, no matter how the operating range changes, the average value of the control coefficient α will remain approximately 1.0 and will be the ideal air-fuel ratio, which is due to the response delay of 6 feedback. It is possible to eliminate transient deterioration of exhaust gas.

Incidentally, the control characteristics of an engine differ greatly from one engine to another due to variations in the characteristics of the engine itself and variations in the characteristics of various sensors and actuators used for control.

For this reason, as for each region correction coefficient Kr required in the above-mentioned region correction method, it would not be a good idea to adopt a method of creating one in advance for the standard engine and using it for all other engines. It doesn't make any sense,
It is necessary to create a ROM for each engine independently and install it as a ROM dedicated to that engine, which reduces productivity and tends to increase costs.

In addition, the characteristics of the engine itself, sensors, and actuators mentioned above change over time, so even if the area correction coefficient is set at the time of manufacture, it will become meaningless after a certain period of time. I often put it away.

Therefore, a nonvolatile memory in which data can be written and rewritten is used to store the region correction coefficient Kr, and the region correction coefficient Kr is sequentially written and replenished for each operating region by learning during engine operation. In recent years, a learning control method has been attracting attention in which air-fuel ratio control is performed by constantly preparing accurate region correction coefficients Kr based on the latest operating results. Ta.

Publications JP-A-54-20231 and JP-A-54-57
The basic idea is shown in No. 029.

According to this learning control method, there is no need to prepare area correction coefficients first, and when there is a change in engine characteristics, the area correction coefficients are self-corrected accordingly. Correct control can be expected, and deterioration of exhaust gas, including during transient conditions, can be prevented.

[Purpose of the invention]

The present invention has been made in view of the above-mentioned circumstances, and its purpose is to supplement area correction coefficients by learning,
It is an object of the present invention to provide an air-fuel ratio control method in which changes are made rationally and learning results are sufficiently reflected in control by a relatively simple program configuration.

[Summary of the invention]

To achieve this objective, the present invention provides a memory area for holding area correction coefficients for use in air-fuel ratio control, a memory area for holding new area correction coefficients obtained through learning, and a memory area for holding area correction coefficients for use in controlling the air-fuel ratio. By providing a memory area that holds the area correction coefficients based on the learning results at the time immediately before the learning results are obtained, it is possible to streamline the process of setting new area correction coefficients and updating the area correction coefficients based on the learning results. It is characterized by the following points.

[Embodiments of the invention]

Hereinafter, the air-fuel ratio control method according to the present invention will be explained in detail with reference to the illustrated embodiments.

In one embodiment of the present invention, the hardware configuration and general operation of fuel injection control are almost the same as the conventional example explained in FIG. 1, but some specific controls are different. A part of the control processing by the microcomputer system included in the electronic control device 1 shown in FIG. 1 is different from the conventional example. Therefore, the following explanation will focus on these different points.

In the following explanation, in order to emphasize that the above-mentioned area correction coefficient Kr is based on learning correction, this coefficient Kr will be expressed as Kt, and will hereinafter be referred to as a learning coefficient.

Therefore, in this embodiment, the injection time T of the injector 4
= is expressed as the following equation (4) instead of the above equation (3).

1'j=Niso Tp・α・Kt・ΣKt ・・・・・・・・・
・・・・・・・・・(4) Now, 0. Assuming that the output signal of the sensor 3 is λ, this signal λ outputs a binary signal (a signal that takes either a high level or a low level value) depending on the presence or absence of oxygen in the exhaust gas. Therefore, in order to perform air-fuel ratio control based on this binary signal, as is well known, the output signal λ of the 01 sensor 3 is checked, and it is determined whether the output signal λ is from a high level (air-fuel ratio is rich) to a low level (similarly lean)
The control coefficient α is increased or decreased in a stepwise manner each time the level is reversed or reversed from a low level to a high level, and then it is gradually increased or decreased.

Therefore, FIG. 2 shows how the control coefficient α changes depending on whether the signal λ is rich or lean.

Therefore, we investigated the extreme value of the control coefficient α that appears when the output signal λ of the O1 sensor 3 is inverted, and the extreme value when it changes from lean to rich is αmax s The extreme value when it changes from rich to -7 is αm10, and from these, the average value αave of the coefficient α is determined by the following formula.

The concept of this average value αave is based on, for example, Japanese Patent Application Laid-open No. 5
This method is known from, for example, Japanese Patent No. 7-26229.

Therefore, in the embodiment of the present invention, upper limit values T, U, L and lower limit values T, L, L are set for this average value αave as shown in FIG. 2, and the average value αave becomes T, U, L.
Or when outside the range of T, L, L, this average value αav
The deviation value between e and α=1.0 is extracted and set as the learning coefficient KL. The process of extracting the learning coefficients is carried out in all of the engine operating ranges where feedback control is being performed.

FIG. 3 is an example of a memory map in which the learning coefficient Kt is written. The engine operating range is determined by the basic fuel injection time Tp and the engine rotation of 13N, and the learning coefficient Ktfa is set as described above in accordance with this operating range. '5 each is stored.

Further, one of the conditions for extracting this learning coefficient i<z is as follows. In other words, one of the above conditions is when the extreme value of the control coefficient α appears repeatedly at least n times (n is a predetermined value such as 5, for example) while the engine operating state remains within the same operating range. There is.

The map shown in Fig. 3 is the learning coefficient i used when controlling the fuel injection time TL in a constant manner according to equation (4).
Since this is a map that stores aK, 4, this is hereinafter defined as a steady learning map.

Now, as is clear from this FIT diagram 3, in this example, the basic fuel μ^ ejection time l''p, which corresponds to the engine load as is clear from equation (2), is set to 0.
By dividing into 8 parts from Ta2 to Ta2, and dividing the engine speed N into 8 parts from 0 to N7, 8 x 8
=64 division points are obtained, and these are taken as the engine operating range. In this embodiment, the writing and modification of the learning coefficient Kt for this steady learning map is not done directly with 5, but with a back-up map having the same memory area configuration as this steady learning map, as shown in FIG. This is done using two more maps called comparison maps.

Therefore, a routine for creating a steady learning map using a plurality of maps as described above will be explained with reference to FIG.

First, as shown in Fig. 5 (1), both the steady state learning map and the comparison map are cleared, and when the engine is operated in this state and the value of the learning coefficient KZ in each operating region is equal to , each time, it is sequentially written only to the corresponding area of the backup file. Note that the routine for calculating the learning coefficient Kl at this time will be described later. Further, at this time, the coefficient Kt in the above equation (4) is set to 1.0.

, 1-5, and the engine continues to be operated for a while, the number of operating regions in which the learning coefficient Kt of the buffer map is written gradually increases. However, while 2
The 64 operating ranges provided in these maps have plenty of margin compared to the operating ranges that appear in an actual engine, so it is not possible to simply operate the engine under normal conditions. The learning coefficient IQ that corresponds to all driving areas is not good at all.

Therefore, when the learning coefficient T<zJ' of the buffer map in the state shown in FIG. 5 (1) - the number C of written operating regions reaches the predetermined value LVC, the The six penetrated data are left as they are and transferred to the comparison map.In addition, the value t at this time is the number of operating regions prepared in these maps, 64.
It is determined to be a predetermined value smaller than , and in this case, for example, a value in the range of 20 to 30 is selected.

Next, as shown in FIG. 5 (3) K, based on the six pieces of data written in the buffer map, a predetermined learning coefficient Kt is written in the operating area of the metalwork in the buffer map with reference to it, and the buffer Create the entire map. In the figure, this is D
It is expressed as Then, this data D is transferred to the steady learning map, and then, as shown in FIG. 5(4), data C, which had been transferred to the comparison map, is returned to the buffer map.

As a result, the entire region of the steady learning map has the name learning coefficient r
<z has been stored, so when the state of this fifth color (4) is obtained, the learning coefficient K of this steady learning map
Control of the fuel injection time TL using equation (4) using t is started. Note that, as described above, up to this point, the calculation of equation (4) has been performed with the learning coefficient Kt being a constant of 1.0.

After starting engine control using the steady learning map, each time a new learning coefficient Kt is set as shown in Figure 2 through learning in a predetermined operating range, As shown in the figure, the learning coefficients Kt of the corresponding operating regions of the steady learning map and the buffer map are corrected by new coefficients, and are set as data D' and C', respectively. Then, this new coefficient (in the case of a buffer map, correction kt includes the case where a learning coefficient is newly written to an operating area where no learning coefficient has been written in tJ<). Each time, the control coefficient α is set to 1.0, and the data written in the buffer map (S) is stored in the comparison map and compared with data C, and the difference in coefficients in each area is determined by a predetermined number. g'l
It is checked whether the number m' has been reached, and when the number has reached m, the routine (6) in the same figure is called and the routines from (2) to (4) are not repeated. In other words, this (6) is (
This indicates that processes 2) to (4) ma and σ) are performed sequentially. tg J6, the above am is the above (
It suffices to set a predetermined value smaller than the number l in 2),
For example, ■? 10 and Jl is enough.

Therefore, according to this embodiment, the average value of the control coefficient α is always kept near 1.0 by using the learning coefficient η′
Since fuel ratio control can be performed, it has excellent responsiveness and can sufficiently suppress the deterioration of exhaust gas in transient conditions.+ Also, the time to rewrite the steady state learning map can be determined by comparing the buffer map and comparison map. This can be done in a sufficiently rational manner, making it possible to learn to accurately respond to changes in the characteristics of each part over time, etc., and to stably provide uniform exhaust gas characteristics over a long period of time.

In addition, in this embodiment, in the steady state learning map shown in FIG. 3, in the region where the basic fuel injection time Tp is 197 or more and the engine speed N is N7 or more, Since control is performed using the learning coefficient Kt in each region of the lower row, optimal power correction can always be automatically obtained even when the operating state of the engine enters the power region.

Next, an example of a learning routine for the learning coefficient Kt and a routine for executing the process shown in FIG. 5 will be described with reference to the flowcharts in FIGS. 6 and 7.

The processing according to these flowcharts is repeated at a predetermined period after the engine is started. First, in step 300 in FIG. Step 30
Proceed to step 2. If the result is No, proceed to step 332.

In step 302, it is determined whether or not the (i) of the O sensor has crossed λ-1 (theoretical air-fuel ratio νF=14.7). The process for changing the control coefficient α in the gradual increase part and gradual decrease part in Figure 2) is performed.The result is Y
If es, the process proceeds to step 304 and calculates the average value αaVe shown in equation (3).

In step 306, it is determined whether the average value αave is within the upper and lower limits shown in FIG.
If s, normal feedback control is being performed, so the counter is cleared in step 326 and the process proceeds to step 332.

On the other hand, if the average value αave is outside the upper and lower limits, in step 308 the difference between the average value αave and 1 is determined by the learning correction function I.
(assumed to be t.Next, in step 310, the current operating range determined from the basic fuel injection time Tp and engine speed N, shown in FIG. Determine whether the operating area has changed by comparing it with the operating area. If the operating area has changed (
Yes) Since there is no operating region in which to write the learning correction amount Kt, the process proceeds to step 326. If the operating range has not changed, the counter is incremented in step 314, and it is determined in step 316 whether the counter has reached n. If the counter value is not n”t (No), proceed to step 332! If the counter value reaches n (Yes), step 3
The counter is cleared in step 18 and the process proceeds to step 320.

In step 320, steps (2) to (4) explained in FIG.
), it is determined whether or not the first steady learning map has been created. If the map has not yet been created, the process proceeds to step 322 and subsequent steps, and the operation (1) described in FIG. 5 is performed. In step 322, it is determined whether a coefficient inverse has already been written in that operating region. If it has already been written (Yeg), do nothing and proceed to step 332. If the result is No, in step 324 the learning correction amount Kt calculated in step 308 is written into the operating region. Step 3
If you created the first steady learning map in 20 (Yes
) Proceed to step 328 and subsequent steps (5) as explained in FIG.
, perform the operation (6) 5. In step 328, the learning correction amount Kt is applied to the division points of the steady learning map and the buffer map.
Add. Then, in step 330, the air-fuel ratio correction coefficient is set to 1.0.

Therefore, the processing according to steps 300 to 332 is repeated, so that the i explained in FIG.
This means that the operations (5) and (6) are obtained.

Next, the operations (2), (3), and (4) explained in FIG. 5 will be explained using the flowchart in FIG. 7.

In step 350, it is determined whether the first steady learning map has been created. If not created yet (NO) Step 35
Proceed to step 4 and check the number of writes in the pack. When the number reaches 2, the process proceeds to step 356, but if the number does not reach 2, the process proceeds to step 370. If you have created the first steady learning map in step 350 (Yes
) At step 352, the difference in data between the buffer map and the comparison map is checked. If there is a difference of 1cm in content between the buffer map and the comparison map, step 35
Proceed to step 6 to create a steady learning map. m in its contents
If there is no difference, proceed to step 370.

At step 356, a map creation flag is set to prohibit writing of learning results. Step 358 transfers the contents of the buffer map to the comparison map, and step 36
In step 0, a steady learning map is created using a buffer map. In step 362, the contents of the created back-up map are transferred to the steady learning map, and in step 364, the contents of the comparison map are transferred to the buffer map. Step 3
At step 66, a flag indicating that the steady learning map has been created is set. This flag is used for determination at step 350 and step 320 in FIG. In step 368,
The map creation flag set in step 356 is reset.

Next, FIG. 8 shows the operation of another embodiment of the present invention. This embodiment differs from the embodiment shown in FIG. 2 not in the average value of the air-fuel ratio control coefficient α, but in its The instantaneous values are T, U,
The learning coefficient is calculated when L (upper limit) or T, L, L (lower limit) is exceeded.
, U, L, or the amount Kt below T, L, L is set as Δα, and Δα is the learning coefficient, and the process is as shown in the flowchart in Figure 9. It becomes like this.

In the following embodiments, it is assumed that all learning coefficients i<z written in the steady learning map are below a value that can be written in this map. However, if the change in the characteristics of each part becomes larger than that, the learning coefficient used to correct it will have a large negative value (and will exceed the limit value that can be written in each area of the steady learning map). Therefore, if even one value of the learning section fJ, kKt reaches the limit value that can be written to the map,
By increasing or decreasing a certain value in the entire area of the map, the average value of the entire area should be brought closer to the reference value, 1, or 0, and the increased or decreased value should be included in the coefficient of equation (4). By doing so, it is possible to shift the numerical values of the entire map, and with this, even large changes in red time can be absorbed (and sufficient correction can be made by one step).

Next, still another embodiment of the present invention is shown in FIG.
This will be explained with reference to FIG.

In this embodiment, in a large transient control state such as during engine acceleration or deceleration, in addition to the learning coefficient Kt described above, an independent correction coefficient is used. ), transient states such as acceleration and deceleration of the engine can be known from the rate of change ΔTp of the basic fuel injection time Tp per unit time. Then, during the acceleration period t and deceleration period t at this time, the air-fuel ratio control coefficient α takes extreme values a and b, as shown in FIG.

Therefore, these extreme values a and b are set to predetermined upper limit values (K, U,
L) or below the predetermined lower limit values (LL, L), calculate the difference between them KaCC and Kdec, and calculate these as the acceleration learning correction amount Kac and the deceleration learning correction amount Kd.
Let it be ec. Then, as shown in FIGS. 11 and 12, the operating range is determined by taking the change rate ΔTp of the basic fuel injection time Tp in the horizontal direction and the engine speed N in the vertical direction as in the steady learning map described above. The information is written in the corresponding driving regions of the divided acceleration learning map (FIG. 11) and deceleration learning map (FIG. 12).

On the other hand, in accordance with this, in this embodiment, the injection time TA by the injector 3 is calculated and controlled according to the following equation.

φTp・α・(Kt+Kt)eΣKb−・−− in TL−
−−−−−+5) Here, Kt is the transient learning coefficient, and when the transient state is due to acceleration, the acceleration learning correction amount Ka read from the corresponding driving region of the acceleration learning map
00 is used as this coefficient Kt, and during deceleration, the deceleration learning correction amount Kdec read from the corresponding driving region of the deceleration learning map is used as the coefficient Kt.

Therefore, according to this embodiment, when the operating state of the engine is changing relatively slowly, the learning coefficients read from the corresponding operating region of the steady learning map are As a result, appropriate control is performed for each operating region, and when the engine is in a transient operating state, in addition to the control using the learning coefficient t<z in the steady state described above, additional control is performed based on the details of the transient state. Accordingly, the acceleration learning correction amount KaCC or deceleration learning correction amount K is extracted from the acceleration learning map or deceleration learning map.
Further fine control is performed by the dec, so that no matter what the operating state is, appropriate air-fuel ratio control can always be performed, and exhaust gas can always be kept in the best condition.

Next, an example of the learning routine for the acceleration learning correction amount KaCC and the deceleration learning correction amount Kdec in this embodiment will be explained as follows.
This will be explained using 74-chard in Figure 3.

In step 400, it is determined whether or not feedback control is in progress. If it is not under control, proceed to step 424. If the control is in progress, proceed to step 402 and check whether the outputs of the six sensors have reversed, and if the output has just been reversed, proceed to step 404.
Sentence forward. If it is not immediately after reversal, the process proceeds to step 424.

In step 404, acceleration or deceleration is checked. Incidentally, the acceleration/deceleration may be checked by looking at the change in the basic fuel injection time Tp per certain period of time. If it is not acceleration or deceleration, the process advances to step 424. Step 406 for acceleration/deceleration
Proceed to.

In step 406, a steady learning map is created (tf
FJll-Noh W9 h) Fuh> or Leopard 1 Chu Nasu If the used Misakidupu has not been created, proceed to step 424. If the steady learning map is in a use permitted state, the process advances to step 408. In step 408, the air-fuel ratio control coefficient α is determined as shown in FIG.
)) is within the upper and lower limit values.

If it is within the upper and lower limit values, the process proceeds to step 424. The result is N
If o, the process advances to step 410. In step 410, if the air-fuel ratio control coefficient α is above the upper limit value (K, U, L), the process proceeds to step 412; if the result is No, the process proceeds to step 4.
Proceed to step 14 to calculate learning correction ■Δα for each acceleration/deceleration. In the next step 416, an operating range is calculated based on the engine rotational speed N at the time when acceleration/deceleration is detected and the rate of change ΔTp of the basic fuel injection time at that time. In step 418, it is determined whether the detected acceleration or deceleration is acceleration or deceleration. If acceleration, the acceleration learning correction amount Δα is added to the acceleration learning map in step 420, and if it is deceleration, the deceleration learning map Δα is added to the deceleration learning map in step 422. Add.

Here, the learning correction amount of acceleration/deceleration is not limited to Kaco or Kdec as shown in FIG. 10(b), but 1.
If it is determined by the deviation from 0, step 41
There is no need to separate 2 and 414, and the learning correction amount can be obtained using the following equation.

Δα=α−1−・・・・・・・・・・・・・・・ (6
) Also, the rate of change ΔTp in the basic fuel injection time is not limited to this, and may also be changes in intake negative pressure, throttle opening, intake air flow rate, etc. In this case, acceleration/deceleration may be adjusted accordingly. Learning map (Figures 11 and 12)
It is obvious that this is caused by changes in engine speed and suction negative pressure.

〔Effect of the invention〕

As explained above, according to the present invention, it is possible to calculate the learning coefficient and to rationally create and modify the map that stores it, so it is possible to fully utilize the features of the learning control method and Even if characteristics of the various actuators and sensors required for fuel ratio control vary or change over time, the operating conditions are always automatically corrected, making it possible to maintain exhaust gas conditions at normal pressure.

Furthermore, according to the present invention, since correction is performed using the steady learning map even in the power range where air-fuel ratio feedback control is not performed, the effects of variations in characteristics of actuators, sensors, etc. and changes over time are prevented even in this range. This allows for optimal power correction at all times.

[Brief explanation of the drawing]

FIG. 1 is a schematic configuration diagram showing an example of an air-fuel ratio feedback type engine control device, FIG. 2 is an explanatory diagram showing the operation of an embodiment of the present invention, and FIG. 3 is a diagram of a steady-state learning map used in the present invention. FIG. 4 is a conceptual diagram of map combination in the present invention, FIG. 5 is an explanatory diagram of map creation 1i1+ creation in the present invention, and FIGS. 6 and 7 are similar diagrams showing map creation processing. FIG. 8 is an explanatory diagram showing the operation of another embodiment of the present invention, FIG. 9 is a flowchart also for explaining the operation, and FIG. Explanatory diagram showing, No. 11
12 and 12 are conceptual diagrams of the map, and FIG. 13 is a flowchart for explaining the operation. 0! Sensor, 4...Injector. 1st figure 211 0.951 31f 411 uf! ! J4alh* Dimension rr (ms)p Fig. 5 117 Fig. 8 1111 @Zan ≧ Hyakushu): Heka 1 Crane

Claims (1)

[Scope of Claims] 1. A correction coefficient that divides the operating state of the engine into a plurality of operating regions according to operating items such as its rotational speed and load size, and that is learned and set for each of these operating regions. In an air-fuel ratio control method for an internal combustion engine that performs control based on the above, means is provided for storing the correction coefficients as a plurality of independent storage data groups, and these storage data groups are used for the control. a first group consisting of correction coefficients, a second group consisting of correction coefficients that are sequentially changed according to the results of learning based on feedback control, and a
and a third group of correction coefficients for determining when to rewrite the correction coefficients of the first group using i positive coefficients. 2. Claim 1 is characterized in that a plurality of operating regions having different operating items are provided, and control is performed by simultaneously using correction coefficients set for each of the operating regions. Air-fuel ratio control method.