JPH0615832B2 - Air-fuel ratio controller - Google Patents

Description

TECHNICAL FIELD The present invention relates to an air-fuel ratio control device.

(Prior Art) Recently, in order to meet various requirements such as engine output improvement, fuel consumption, and exhaust emission countermeasures, the concept of learning control is introduced and the air-fuel ratio tends to be controlled more precisely. In such learning control, when the learning value is stored, the number of divisions of the operating region (hereinafter referred to as the number of region divisions) and the selection of division parameters are considered.

As a conventional air-fuel ratio control device of this type, for example, JP-A-57-188745 and JP-A-57-1432 are known.
The one described in Japanese Patent No. 34 is known. Each of these devices uses an air-fuel ratio correction coefficient α that corrects the air-fuel ratio to the stoichiometric air-fuel ratio based on the output of an oxygen sensor provided in the exhaust passage.
₁ and feedback control of the air-fuel ratio, while α ₁
= 1, that is, when the control loop is open, the difference from the air-fuel ratio (hereinafter referred to as the open air-fuel ratio) is sequentially learned, and the optimum value is stored as the learning value α ₂ , When the output of the oxygen sensor has a response delay or the output is unstable, such as when the engine is started or when the engine is started, the learning value α ₂ is used as a learning correction coefficient and the air-fuel ratio is set to a predetermined air-fuel ratio in an open loop. By controlling, responsiveness and controllability at the time of starting are improved.

Here, when the learning value α ₂ is stored, the former device uses the intake air amount Qa as a division parameter of the operating region and divides the intake air amount Qa into 16 as shown in FIG. The learning value α ₂ is assigned to each divided block. On the other hand, in the latter device, as shown in FIG. 13, the intake air amount Qa and the rotation speed N are used as the division parameters, and the intake air amount Qa is divided into 32 and the rotation speed N is increased.
Is divided into a large number of areas every 200 rpm, and the learning value α ₂ is assigned to the two-dimensional table map of Qa and N.

However, in such a conventional air-fuel ratio control device, in the former case, since the split parameter is only the intake air amount Qa, for example, the learning value cannot be corrected based on the variation in the injection characteristics of the injectors, and the air-fuel ratio control is not possible. Can not be expected to improve the accuracy of. On the other hand, in the case of the latter, since it is a two-dimensional map in which the intake air amount Qa and the rotation speed N are the division parameters,
Although the accuracy of the air-fuel ratio control can be improved compared to the former, for example, in an engine with a maximum rotation speed of 6000 rbm, the divided area of the rotation speed N becomes 30 blocks and finally
A memory having a large number of addresses of 960 points (960 = 32 divisions x 30 blocks) is required, which causes the device to become complicated and costly. Therefore, for example, it is conceivable to reduce the number of area divisions, but in this case, the following problems occur.

(I) Deterioration of drivability FIG. 14 is a learning value map when the number of area divisions is reduced as an example. In this map, for example, when the data of the Qa ₃ region is plotted on the drawing according to the rotation speed N,
As shown in FIG. 15, it can be seen that the learning value α ₂ has a step at the boundary of each region. Therefore, when the rotation speed N changes near the boundary of the region, the value of the learning value α ₂ suddenly changes and the air-fuel ratio also suddenly changes. As a result, the drivability deteriorates because the torque suddenly changes and a shock is generated in the vehicle. Therefore, it is unavoidable to increase the number of region divisions within a range in which deterioration of drivability can be avoided.

(II) Reduction in learning accuracy of learning value α ₂ FIG. 16 shows the relationship between the intake air amount Qa and the open air-fuel ratio when the rotation speed N is constant. As shown in the figure, the open air-fuel ratio greatly changes depending on the value of Qa. Therefore, at this time, the area division of Qa is Qam _-1 across Qam.
And Qam ₊₁ as the minimum unit, the learning value α ₂
Is an average value in each region and cannot be said to accurately correspond to the actual intake air amount Qa, and the learning accuracy thereof decreases. For this reason, it is necessary to subdivide the divided area as in the above (I). It should be noted that other factors that cause variations in the open air-fuel ratio include variations in the injection characteristics of the injector. However, this is not considered.

(Object of the invention) Therefore, the present invention divides the operating region into a plurality of predetermined interpolation regions, stores the learned value in the index address dividing the learning interpolation region by a predetermined storage interpolation operation, and performs a predetermined reading. By reading this learning value by interpolation calculation, the number of region divisions is reduced to reduce the number of memories to avoid complication of the device and high cost, while maintaining high accuracy of the learning value and improving the accuracy of air-fuel ratio control. The purpose is to improve.

(Structure of the Invention) FIG. 1 is an overall structure diagram for clarifying the present invention.

a is an oxygen sensor for detecting the oxygen concentration in the exhaust gas, b is an operating state detecting means for detecting the operating state of the engine, and c is an air-fuel ratio correction coefficient for correcting the air-fuel ratio to a predetermined air-fuel ratio based on the output of the oxygen sensor. A correction coefficient calculation means for calculating, a learning value calculation means for calculating a learning correction coefficient for matching the air-fuel ratio with the target air-fuel ratio from the value of the air-fuel ratio correction coefficient, and e for two-dimensionally calculating two parameters corresponding to the operating state. A table map in which the X-axis and the Y-axis of the map are made to correspond to each other, and the coordinates of the intersections of the X-axis and the Y-axis are used as index addresses, and f is based on two parameters that represent the operating state at an arbitrary time point. Two X-axis addresses and Y each having a close value
An index address selecting means for selecting an axis address and selecting four index addresses from the combination of the X-axis address and the Y-axis address, g is a representative one of the four index addresses selected. , An interpolation coefficient calculation means for calculating an interpolation coefficient indicating how far the X-axis address and the Y-axis address corresponding to the operation state at the arbitrary time point are separated, and h is the four index addresses based on the interpolation coefficient Calculate the weight value for each
Memory value calculation means for calculating the memory value for each index address by reflecting each weight value in the learning correction coefficient, i
Is a storage means for storing the stored value in a corresponding index address, j is a unit operation region including two parameters corresponding to an operation state at an arbitrary time point, and an X-axis and a Y axis for dividing the unit operation region are selected. Readout means for obtaining the degree of separation between the axis and the parameter, correcting the stored values of the four index addresses of the unit area based on the degree of separation, and reading the corrected values as learning correction coefficients, k is the air-fuel ratio Supply amount control means for controlling the fuel supply amount so that the air-fuel ratio becomes the target air-fuel ratio on the basis of at least one of the correction coefficient and the learning correction coefficient, and l is the fuel for the engine based on the signal from the supply amount control means. Is a fuel supply means for supplying.

(Example) Hereinafter, the present invention will be described with reference to the drawings.

2 to 6 are views showing an embodiment of the present invention.

First, the configuration will be described. In FIG. 2, reference numeral 1 is an engine, intake air is supplied from an air cleaner 2 to each cylinder through an intake pipe 3, and fuel is injected by an injector (fuel supply means) 4 based on an injection signal Si. To be done. Then, the exhaust gas burned in the cylinder is introduced into the catalytic converter 6 through the exhaust pipe 5, and in the catalytic converter 6, harmful components (CO, HC, NOx) in the exhaust gas are cleaned and discharged by the three-way catalyst. The flow rate Qa of the intake air is detected by the air flow meter 7 and controlled by the throttle valve 8 in the intake pipe 3. The air flow meter 7 outputs a signal Sa having an analog voltage according to the intake air amount Qa. Engine 1
The crank angle Ca is detected by the crank angle sensor 9, and the temperature Tw of the cooling water flowing through the water jacket is detected by the water temperature sensor 10. Further, the oxygen concentration in the exhaust gas is detected by the oxygen sensor 11, and as the oxygen sensor 11, for example, a sensor having a characteristic that the output voltage Vs thereof suddenly changes at the stoichiometric air-fuel ratio is used. The air flow meter 7, the crank angle sensor 9 and the water temperature sensor 10 constitute an operating state detecting means 12, and signals from the operating state detecting means 12 and the oxygen sensor 11 are input to the control unit 13.

The control unit 13 has a function as a correction coefficient calculation means, a learning value calculation means, an address selection means, an interpolation coefficient calculation means, a storage value calculation means, a storage means, a reading means, and a supply amount control means. As shown in detail in the figure, it comprises a CPU 21, a ROM 22, a RAM 23, a RAM 24, an I / O port 25, an A / D conversion circuit 26 and a counter 27. The A / D conversion circuit 26 converts the signals Sa, Tw, Vs input as analog signals into digital signals and outputs them to the CPU 21. The crank angle signal Ca from the crank angle sensor 9 is input to the counter 27, and the counter 27 counts the crank angle signal Ca (for example, 2 ° signal) to calculate the rotation speed N of the engine 1 and the CPU 21. Output. The CPU 21 takes in necessary external data according to the program written in the ROM 22,
In addition, data is exchanged with the RAMs 23 and 24, arithmetic processing is performed, and the processed data is input / output as necessary.
Output to port 25. The crank angle signal Ca from the crank angle sensor 9 is further input to the I / O port 25, and the I / O port 25 receives data and signal Ca from the CPU 21.
And outputs the injection signal Si to the injector 4.
The ROM 22 stores a calculation program in the CPU 21, and the RAMs 23 and 24 store data used for calculation in the form of table map. Although the storage contents of the RAM 23 are lost after the engine 1 is stopped, the RAM 24 is constituted by, for example, a non-volatile memory and retains the storage contents (learned value etc.) even after the engine 1 is stopped.

Next, the operation will be described.

In general, learning control can supplement the control responsiveness, which is a problem of feedback control, and has the advantage that the same control accuracy as that of feedback control can be ensured particularly when open loop control is performed. However, it is difficult to take advantage of such an advantage when the accuracy of the learning value itself is low. Therefore, while increasing the number of area divisions,
According to this, a method of increasing the number of memories is used, but this leads to the complexity of the device and the cost increase.

Therefore, in this embodiment, if the driving state and the learning value α ₂ are accurately correlated in both the storing and reading processes,
It is not always necessary to make these correspond one-to-one in a large number of operating areas through many memories, and according to the method of interpolation calculation, even if the number of reference points is reduced, the values between the reference points are substantially accurate. Note that it is possible to calculate the value corresponding to each reference point, even if it is a value between the reference points. By adopting different interpolation operations, the accuracy of the learning value α ₂ is increased while reducing the number of required memories.

4 and 5 figure is a flowchart showing an air-fuel ratio control of the program written in the ROM 22, reference numeral P ₁ to P
₂₈ indicates each step of the flowchart. This program is executed once for each engine revolution, for example.

FIG. 4 is a flowchart showing a main routine of air-fuel ratio control. First, at P ₁ , the output signal Sa of the air flow meter 7 is A / D converted into a digital signal Da, and at P ₂ , this digital signal Da is predetermined Da-Q.
The intake air amount Qa is converted using an a map (not shown).
This is because the digital signal Da and the intake air amount Qa are not in a simple proportional relationship. Then, at P ₃ , the rotation speed N
Is read and the basic injection amount Tp is calculated in P ₄ according to the following equation.

Tp = K · Qa / N, where K: a constant, and then at P ₅ , the final injection amount Ti is calculated according to the following equation, and at P ₆ , the injection signal Si having the pulse width corresponding to the final injection amount Ti is I / Set to O port 25.

Ti = Tp × COEF × α ₁ × α ₁ + TS In the formula, COEF is various increase factors, and for example, various increase corrections (including decrease corrections) of the basic injection amount Tp are performed based on the cooling water temperature Tw, acceleration increase, and the like. To do. α ₁ is an air-fuel ratio correction coefficient at the time of feedback control calculated in a subroutine described later, and α ₂ is this air-fuel ratio correction coefficient α ₁
Is a learning correction coefficient for matching the air-fuel ratio with the theoretical air-fuel ratio from the value of. The value of the learning correction coefficient α ₂ is obtained as a learning value when the learning correction is performed so that the air-fuel ratio = theoretical air-fuel ratio based on the value of the air-fuel ratio correction coefficient α ₁ . Hereinafter, the learning value is set to α ₂ for convenience of explanation. Then, in the present embodiment, the calculation method in which these α ₁ and α ₂ are used together is adopted by this equation, but in the case of open loop control, for example, α ₁ = 1 is actually set to correct only α _2. Becomes
Note that Ts is a coefficient for correcting the response delay (dead time) of the injector 4. Therefore, the final injection amount Ti of fuel is injected from the injector 4 into the intake pipe 3, and the air-fuel ratio of the intake air-fuel mixture is constantly controlled to the target value, as will be described later.

FIG. 5 is a flowchart showing a subroutine SUB-1 for calculating the air-fuel ratio correction coefficient α ₁ and the learning value α ₂ . At P ₁₁ , the output Vs of the oxygen sensor 11 is A / D converted into a digital signal, and at P ₁₂ , it is determined whether or not the open loop condition is satisfied. The open loop condition is satisfied, for example, when the oxygen sensor 11 is not sufficiently activated at the time of starting the engine or when the response delay of the oxygen sensor 11 cannot be ignored during the transient operation. When the open loop condition is satisfied, the process proceeds to P _13, it executes the open loop control. On the other hand, when the condition is not satisfied, P ₁₄
In the following steps, feedback control is executed, and α ₂ is learned from the value of α ₁ at this time.

First, the time of feedback control will be described. First,
An air-fuel ratio correction coefficient α for correcting the air-fuel ratio to the stoichiometric air-fuel ratio at P _14.
Calculate the value of ₁ . This calculation is performed as follows, for example. The digital conversion value of the oxygen sensor output Vs is compared with a comparison reference value S / L (S / L: a value approximately midway between the upper limit and the lower limit when the output Vs suddenly changes at the stoichiometric air-fuel ratio), and Vs <S /
When L, it is judged to be leaner than the stoichiometric air-fuel ratio, and the value of the air-fuel ratio correction coefficient α ₁ for correcting the air-fuel ratio to the stoichiometric air-fuel ratio is increased. On the other hand, when Vs> S / L, it is determined that the air-fuel ratio is richer than the stoichiometric air-fuel ratio, and the value of the air-fuel ratio correction coefficient α ₁ is decreased. The increase / decrease of the air-fuel ratio correction coefficient α ₁ is performed by PI (proportional integral) control. As a result, the value of the air-fuel ratio correction coefficient α ₁ is appropriately corrected according to the operating state, and the air-fuel ratio is accurately controlled to the stoichiometric air-fuel ratio. Next, at P ₁₅ , it is determined whether or not the values of various increase factors COEF are COEF = 1. When COEF = 1, the process proceeds to P ₁₆ to learn α ₂ from the value of α ₁ and COEF ≠ 1. proceeds to P ₁₃ it is determined not to perform the learning time. This is because if COEF ≠ 1, the final injection amount Ti when being maintained at the stoichiometric air-fuel ratio is corrected not only by α ₁ but also by COEF. Therefore, in this embodiment, α ₂ is learned from the value of α ₁ only when COEF = 1. In addition, COEF
Not limited to learning under the condition of = 1, α ₂ may be learned from the value of [COEF × α ₁ ] including COEF. At P ₁₆ , the value of α ₂ corresponding to the operating state at that time is calculated from the value of α ₁ . This is, for example, α ₁
Is sampled a predetermined number of times to obtain the average value thereof, and α ₂ is calculated as a learning value from this. Then, at P ₁₇ , the learning value α
₂ is stored (details will be described later in subroutine SUB-2) and the process proceeds to P ₁₃ .

On the other hand, when the open loop condition is satisfied at P ₁₂ , the learning value α ₂ according to the operating state is read at P ₁₃ . For convenience of description, this reading method will be described after the method of storing the learning value α ₂ is described. Therefore, in the main routine, the final injection amount Ti is determined according to the above equation, and the air-fuel ratio is controlled in the open loop to the stoichiometric air-fuel ratio. In this case, since the learning value α ₂ is highly accurate, the oxygen sensor 11
Of the output Vs is not stable, or during a transient operation in which the response delay of the output Vs cannot be ignored, the air-fuel ratio can be controlled to the target value with the same high accuracy as in the feedback control. .

FIG. 6 shows a subroutine SUB- for storing the learning value α ₂ .
6 is a flowchart showing No. 2.

First, the principle of this store system will be described. In the present embodiment, the rotation speed N and the basic injection amount Tp are used as the division parameters of the operating region, and the operating region is divided into a grid by the parameters N and Tp as shown in FIG. The number of sections is, for example, the latter example (960
It is set to a much smaller number than
The amount of memory required is small. Then, the learning value α ₂ is stored as ▲ α ⁿ _2m ▼ at the addresses Nn and Tpm (which refer to the address of the lattice point, hereinafter referred to as the index address) designated by the division numbers of N and Tp. Therefore,
For example, the number of divisions (the number of grids) of the N axis (corresponding to the X axis described in the gist of the invention) is No, and the number of divisions (the number of grids) of the Tp axis (corresponding to the Y axis described in the gist of the invention) is Mo. At this time, the number of grid points is No × Mo, and the same number of learning values α ₂ are calculated and stored. In this embodiment, since the number of index addresses Nn and Tpm is smaller in the present embodiment than in the latter example, the accuracy of the learning value α ₂ may decrease if this is left as it is. That is, the learning value ▲ α ⁿ _2m ▼ is calculated as a simple average value within the unit section, and this is calculated as the index address Nn, T
If the data is simply stored in pm, the learning accuracy of (II) shown in the conventional example is deteriorated.

Therefore, if the present inventor can accurately store the actual driving state represented by N and Tp and the learned value α ₂ and store them, the accuracy of the learned value α ₂ will decrease even if the number of index addresses is small. By paying attention to the fact that it can be prevented, the accuracy of the learning value ▲ α ⁿ _2m ▼ stored in the index addresses Nn and Tpm is increased by applying the principle of interpolation calculation at the time of storing.

In FIG. 6, the learning correction deviation v corresponding to the present N and Tp is calculated at P ₂₁ according to the following equation.

v = W × (α ₁ −1), where W: constant The learning correction deviation v corresponds to the degree of separation of α 1 from ₁ , and the learning correction coefficient is corrected to eliminate v. Becomes In addition, the operating state X at this time, that is, N and T
p (however, Nn ≦ N <Nn _{+ 1} , Tpm ≦ Tp ≦ Tp
m ₊₁ ) is four index addresses (n,
m) (n + 1, m) (n, m + 1) (n + 1, m + 1)
It is located at a point X in the interpolation area Hs divided by. These four index addresses are N, Tp at any time.
Two N-axis values and two Tp-axis values that are closest to the magnitude (corresponding to the two parameters corresponding to the operating state) are selected and calculated from the combination (2 ² ) of these values. In this case, there is no memory area in the interpolation coefficient Hs itself, and the memory area is limited to each of the above index addresses. Therefore, the learning value α ₂ is 4 as it is.
When stored in one index address, all four points have the same value and the accuracy of the learning value α ₂ decreases. Therefore, one index address that is a representative of the four index addresses in P ₂₂ and P ₂₃ . For example, the degree of separation from the index address (n, m) at point X is calculated as the memory interpolation coefficients a and b in the N-axis and Tp-axis directions according to the following equations.

a = (N-Nn) / (Nn _{+ 1} -Nn) ... b = (Tp-Tpm) / (Tpm _{+ 1} -Tpm) ... Therefore, the values of a and b obtained by the above equation are It is a coefficient (interpolation coefficient) showing how far the two parameters (N, Tp) corresponding to the operating state at any time point are separated from one representative index address (n, m). From these stored interpolation coefficients a,
By using b, it is possible to quantitatively represent the positional relationship of X (N, Tp) with respect to each index address, as shown by the broken line arrow in FIG. By using such a quantitative positional relationship, it is possible to accurately separate X (N, Tp) by appropriately weighting each index address. Based on this principle, P _{24 to} P ₂₇
According to the learning value of each index address ▲ α ⁿ _2m ▼, ▲ α ^{n + 1}
_2m ▼, ▲ α ⁿ _{2m + 1} ▼, ▲ α ^{n + 1} _{2m + 1} ▼ are calculated respectively,
Store these at each index address at P ₂₈ .

However, α ₂ old: old data up to the previous time stored in RAM 24 α ₂ new: new data of this time In the above equations, a and b are interpolation coefficients representing the degree of separation, and N, Tp at arbitrary points in time. Becomes larger as it approaches one index address (n, m) that is a representative. In addition, (1-a) and (1-b) are opposite to each other, and the value becomes smaller as N, Tp at any given time approaches one representative index address (n, m). For example, since the formula is an updating formula for the learning correction coefficient stored in the representative index address, in this case, (1-a) and (1-
b) is used, and N, Tp is applied to the learning correction coefficient.
It is possible to give a weight value that becomes larger as is closer to. Therefore, X (N, T
The accuracy of the learning value α ₂ corresponding to p) can be maintained high and stored in each index address. At the same time, learning correction of the learning value α ₂ based on the learning correction deviation v is performed at the same time, so that the learning data always has the latest value, and the reliability of the data is improved.

Next, a method of reading the learning value α ₂ stored in each index address will be described.

When the operating state is X (N, Tp), the unit operating region corresponding to this X is shown in FIG. At this time, X
Index address (n,
m) (n, m + 1) (n + 1, m) (n + 1, m + 1)
Of the learning value alpha ₂ is only simply read, the learning value alpha ₂
Is not exactly correlated to the operating state X. Therefore,
Interpolation calculation is performed as in the store method to accurately correlate the two. First, the unit operation area is taken as an interpolation area Hs, and the X point position in the interpolation area Hs is calculated according to the following equations as the read interpolation coefficients c and d in the N-axis and Tp-axis directions, respectively. To do.

c = (N-Nn) / (Nn _{+ 1} -Nn) ... d = (Tp-Tpm) / (Tpm _{+ 1} -Tpm) ... Next, the interpolation value E on the Tpm axis corresponding to the X point is calculated as follows. Calculate according to the formula.

Similarly, the interpolation value F on the Tpm axis corresponding to the point X is calculated according to the following equation.

That is, the interpolation values E and F are obtained by linear interpolation. Then, the learning value α corresponding to the point X from these interpolation values E and F
₂ is linearly interpolated according to the following equation.

α ₂ = E + d × (FE) ... Therefore, the learning value α ₂ obtained by the formula becomes a value accurately correlated with the operating state (N, Tp) at the X point, and the accuracy of the data is high. can do.

As described above, although the number of memories is small, the learning value α ₂ is stored in and read out from the learning value α _{2 by using} a predetermined interpolation calculation.
_{Since 2} is accurately correlated with the operating state (N, Tp), the accuracy of the learning value α ₂ can be maintained high. As a result, it is possible to improve the accuracy of the air-fuel ratio control using the learning value α ₂ while avoiding complication of the device and high cost.

In the present embodiment, N and Tp are used as the division parameters to perform learning correction based on Tp in order to improve the variation in the injection characteristics of the injectors. However, the present invention is not limited to this, and N and Qa are used, for example. Good. The difference in the adoption of these division parameters appears as the difference in the learning correction range in the engine operating region, and the practical learning correction range is shown by the diagonal lines in FIG. 10 for the former and FIG. 11 for the latter. It will be something like. As is clear from these figures, when N and Tp are used as the division parameters, all the lattice points of the memory can be effectively used. And this is Q
This means that Tp can reduce the number of memories as compared with a.

Further, although the example in which the air-fuel ratio is controlled to the stoichiometric air-fuel ratio has been shown in the present embodiment, the present invention is not limited to this. For example, the value of α ₁ when controlling to the stoichiometric air-fuel ratio is learned, and based on this learned value, it is corrected to another target air-fuel ratio (for example, lean air-fuel ratio A / F = 18) Of course, the present invention can be applied to an example in which the air-fuel ratio correction coefficient is calculated and the target air-fuel ratio is controlled based on this. In short, the learning correction coefficient α ₂ is stored from the value of α ₁ and the so-called learning control method of performing the air-fuel ratio control by using this α ₂ can be applied to all.

Furthermore, it goes without saying that each of the above-described interpolation calculation methods is only an example for achieving the object of the present invention, and various modifications can be made within the scope of the object.

Further, the present invention is not limited to electronically controlled fuel injection, but can be applied to, for example, a carburetor type engine.

(Effect) According to the present invention, it is possible to maintain the accuracy of the learning value at a high level while reducing the number of regions and the number of memories.
The accuracy of the air-fuel ratio control can be improved while avoiding the complexity of the device and the high cost.

[Brief description of drawings]

FIG. 1 is an overall configuration diagram of the present invention, FIGS. 2 to 11 are diagrams showing an embodiment of the present invention, and FIG. 2 is a schematic configuration diagram thereof.
FIG. 4 is a circuit configuration diagram of the control unit, FIG. 4 is a flowchart showing a main routine of the air-fuel ratio control, and FIG. 5 is an air-fuel ratio correction coefficient α ₁ and a learning value α ₂
FIG. 6 is a flowchart showing a subroutine SUB-1 for calculating the following, FIG. 6 is a flowchart showing a subroutine SUB-2 for storing the learning value α ₂ thereof, and FIG. 7 is an operating region with the rotation speed N and the basic injection amount Tp as parameters. , And FIG. 8 is a diagram showing a method of storing the learning value α ₂ .
FIG. 9 is a diagram showing a method of reading the learning value α ₂ ,
FIG. 11 is a diagram showing a learning correction range of the operating region with the rotational speed N and the basic injection amount Tp as parameters, and FIG. 11 shows a learning correction range of the operating region with the rotational speed N and the intake air amount Qa as parameters. FIGS. 12 to 16 are views showing a conventional air-fuel ratio control device, FIG. 12 is a map showing the relationship between the intake air amount Qa and the learning value α _2, and FIG. 13 is its rotational speed N and A map showing the relationship between the intake air amount Qa and the learning value α ₂ ,
FIG. 14 shows the rotational speed N, the intake air amount Qa, and the learning value α.
₂ is a map showing the relationship with ₂ by numbers, FIG. 15 is a view showing the map of FIG. 14 by the relationship between the rotational speed N and the learning value α _2, and FIG. 16 is the intake air amount Qa and the open air-fuel ratio. It is a figure which shows the relationship of. 1 ... Engine, 4 ... Injector (fuel supply means), 12 ... Operating state detection means, 13 ... Control unit (correction coefficient calculation means, learning value calculation means, address selection means, interpolation coefficient calculation means, stored value Computing means, storage means, reading means, supply amount control means).

 ─────────────────────────────────────────────────── ─── Continuation of front page (56) Reference JP-A-60-153448 (JP, A) JP-A-58-107875 (JP, A) JP-A-56-151267 (JP, A) JP-A-57- 18440 (JP, A) JP 59-138753 (JP, A) JP 58-53184 (JP, B2)

Claims (1)

[Claims] 1. An a) oxygen sensor for detecting the oxygen concentration in exhaust gas, b) an operating state detecting means for detecting the operating state of an engine, and c) an air-fuel ratio set to a predetermined air-fuel ratio based on the output of the oxygen sensor. Correction coefficient calculation means for calculating an air-fuel ratio correction coefficient to be corrected; d) learning value calculation means for calculating a learning correction coefficient for matching the air-fuel ratio with the target air-fuel ratio from the value of the air-fuel ratio correction coefficient; And a table map in which the coordinates of the intersections of the X-axis and the Y-axis are used as index addresses, and the operating state at any time Based on the two parameters shown, two X-axis addresses and two Y-axis addresses each having a value close to the parameters are selected, and four index addresses are selected from the combination of the X-axis address and the Y-axis address. Index address selecting means, and g) How far apart is the X-axis address and the Y-axis address corresponding to the operating state at the arbitrary time from one representative index address of the selected four index addresses. Interpolation coefficient calculating means for calculating an interpolation coefficient indicating whether or not each of the four index addresses is calculated based on the interpolation coefficient, and each weight value is reflected in the learning correction coefficient. A storage value calculating means for calculating a storage value for each index address, i) storage means for storing the storage value at a corresponding index address, and j) two parameters corresponding to an operating state at an arbitrary time point. Select the unit operation area to be determined, and obtain the degree of separation between the X-axis and Y-axis and the parameter that divide the unit operation area
Reading means for correcting the stored values of the four index addresses of the unit area based on the degree of separation and reading the corrected values as learning correction coefficients; and k) at least one of the air-fuel ratio correction coefficient or the learning correction coefficient.
Supply amount control means for controlling the fuel supply amount so that the air-fuel ratio becomes the target air-fuel ratio on the basis of one or more, and 1) fuel supply means for supplying fuel to the engine based on a signal from the supply amount control means, An air-fuel ratio control device comprising: