JPH0510235A - Ignition timing controller for internal combustion engine - Google Patents

Abstract

PURPOSE:To always set the optimum ignition timing of an internal combustion engine in all the operation regions on the basis of the result of the knocking judgement, by representing the index having vagueness by the membership function and setting the ignition timing by using the fuzzy inference. CONSTITUTION:The first control quanti,ty is the value which is always learned according to the state of an internal combustion engine 1, and the method for setting the renewal quantity is set by using the fuzzy inference. The index having the vagueness such that if the number of engine revolution obtained from the detection signal of a turning angle sensor 5 is in a high revolution region or low revolution region is represented by the membership function. Further, for the number of engine revolution of 0-2000-4000rpm, a triangle having 2000rpm as apex is formed, and the magnitude of the number of engine revolution having vagueness is represented.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an ignition timing control device for an internal combustion engine, which detects knocking occurring in the internal combustion engine and controls the ignition timing of the internal combustion engine.

[0002]

2. Description of the Related Art Generally, in an internal combustion engine, an optimum ignition timing is calculated from a rotational speed of the internal combustion engine (hereinafter referred to as an engine rotational speed) and an intake air amount, and the actual ignition timing is controlled based on the calculated value. is doing. Further, at this time, when knocking of the internal combustion engine is detected by the knock sensor and it is determined that knocking has occurred in the internal combustion engine, the ignition timing is gradually set to the retard side from the optimum ignition timing, and if knocking does not occur. Feedback control is performed such that when the judgment is made, the optimum ignition timing is gradually approached.

However, in the high engine speed range, vibration (noise) of the internal combustion engine itself becomes large, and it is not possible to accurately determine whether knocking has occurred. Therefore, for this reason, the knocking determination result is not reflected in the ignition timing control in the high engine speed region above the predetermined engine speed, that is, the feedback control is stopped and the optimum ignition timing is retarded with the maximum retardation amount. Control to the side.

As described above, since the ignition timing is always set to the maximum retard side in the high speed region, the driver demands a large output of the internal combustion engine when, for example, sudden acceleration is desired in the high speed region. Despite this, there was a problem that a large output could not be obtained in the high rotation range.

Therefore, in order to solve the above-mentioned problems, in the high engine speed range above a predetermined engine speed, the maximum value of the ignition timing retard amount obtained in consideration of the knocking determination result is used. A method has been proposed in which the ignition timing is set to a constant maximum retard side so that a high output can be obtained in a high rotation range by setting the ignition timing of Kaisho 60-478
No. 72).

[0006]

However, the above-mentioned 2
One of the problems in one of the methods is whether it is currently in a high rotation region or in a low rotation region, in other words, whether it is equal to or higher than a predetermined engine speed,
Generally, the determination is made by comparison with a certain threshold value.

More specifically, for example, in the latter case, assuming that the threshold value is the engine speed of 5000 rpm (rpm), if the current speed is 4990 rpm, it is considered that it is not in the high rotation range, and the retard amount is obtained based on the knocking determination result and fed back. Ignition timing is determined by control, only 20r
At 5010 rpm, which is higher than pm, the ignition timing is determined using the maximum value of the retard amount of the ignition timing obtained in the knocking discriminable region as the high rotation region.

That is, the accuracy of knocking determination does not suddenly deteriorate when the engine speed is around 5000 rpm, and it is difficult to accurately determine knocking even when the engine speed is 4990 rpm, for example. However, since the ignition timing is determined by obtaining the retard amount based on the knocking determination result, there is a possibility that the optimum ignition timing cannot be set.

On the contrary, the engine speed is 5010 rp
In m, the optimum ignition timing is set even in such a case, for example, the same ignition timing is set even though the knocking detection accuracy is not deteriorated as compared with the operating state where the engine speed is 7,000 rpm. It cannot be set. In other words, it has not been possible to effectively set the ignition timing in all operating regions by the conventional methods.

Therefore, the present invention has been made in order to solve the above-mentioned problems. For example, an index with ambiguity such as the above-mentioned high rotation region or low rotation region of the engine speed is expressed by a membership function. To provide an ignition timing control device for an internal combustion engine capable of always setting an optimal ignition timing for the internal combustion engine in all operating regions by setting the ignition timing by using fuzzy inference. With the goal.

[0011]

In order to achieve the above object, an ignition timing control apparatus for an internal combustion engine according to the present invention, as shown in FIG. 1, is an internal combustion engine state detecting means for detecting an operating state of the internal combustion engine, and the internal combustion engine state detecting means. Knock determination means for determining the presence or absence of knocking occurring in the engine, basic ignition timing calculation means for calculating the basic ignition timing of the engine according to the operating state of the internal combustion engine, the detection result of the internal combustion engine state detection means and The determination result of the knock determination means is expressed as a membership function as an input condition, the learning amount is obtained using fuzzy reasoning, and the first control amount for controlling the basic ignition timing is learned based on the learning amount. First control amount learning means, second control amount calculation means for calculating a second control amount for controlling the ignition timing based on the determination result of the knock determination means, and Adopt the technical means of and an ignition timing setting means for setting an ignition timing of the internal combustion engine by correcting depending on the basic ignition timing a first control amount and the second control amount.

[0012]

According to the present invention, the basic ignition timing is calculated according to the operating state of the internal combustion engine. Further, the detection result of the internal combustion engine state detection means and the determination result of the knock determination means are expressed by a membership function as an input condition, a learning amount is obtained using fuzzy reasoning, and the ignition timing is controlled based on the learning amount. The first controlled variable to be learned is learned. Further, a second control amount for controlling the ignition timing is calculated according to the presence or absence of knocking in the internal combustion engine, and the basic ignition timing is corrected according to the first control amount and the second control amount. Thus, the ignition timing of the internal combustion engine is set.

[0013]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described in detail below based on the embodiments shown in the drawings. FIG. 2 is a schematic configuration diagram showing the configuration of the apparatus of this embodiment.

In FIG. 2, reference numeral 1 is an internal combustion engine, and 2 is an intake pipe for introducing intake air introduced from an air cleaner (not shown) into the internal combustion engine 1. 3 is a throttle valve that opens and closes in conjunction with an accelerator pedal (not shown) and controls the amount of intake air introduced into the internal combustion engine 1 (intake air amount Q). 4 is an air flow meter that measures this intake air amount Q. The air flow meter 4 incorporates a potentiometer to generate an electric signal of an analog voltage proportional to the intake air amount Q, and this electric signal is input to an electronic control unit described later.

5 is built in the distributor 6,
A rotation angle sensor for outputting a signal for each predetermined crank angle to obtain a rotation speed Ne of the internal combustion engine 1 (hereinafter, referred to as engine speed), 7 is arranged in a cylinder block of the main body of the internal combustion engine 1, and for example, a cylinder. This is a knock sensor that detects knocking from vibration of the block, and output signals from the rotation angle sensor 5 and the knock sensor 7 are also input to the electronic control unit.

Reference numeral 8 calculates an optimum control amount of the ignition system and the fuel system based on detection signals from the above-mentioned sensors and a sensor (not shown) for detecting the state of the internal combustion engine, and accurately controls an injector, an igniter and the like which will be described later. It is a known electronic control unit (hereinafter referred to as ECU) that outputs a control signal for

Further, the ECU 8 is a well-known C for performing arithmetic processing.
PU8a, read-only storage device ROM8 for storing control programs and control constants necessary for calculation
b, a temporary storage device RAM 8c for temporarily storing operation data during operation of the operation device 8a, and an I / O port 9d for inputting / outputting signals from outside the ECU 8.

Further, the ECU 8 obtains the basic ignition timing from the engine speed Ne and the intake air amount Q (basic ignition timing calculating means), and uses the method described later to determine the maximum retardation amount, the first control amount, and the second control amount. And a first control amount learning means and a second control amount calculation means for calculating the control amount. Further, the basic ignition timing is the maximum retardation amount,
It constitutes an ignition timing setting means for setting the ignition timing of the internal combustion engine 1 by making corrections based on the first control amount and the second control amount.

Reference numeral 9 is an injector for supplying fuel to the internal combustion engine 1 at an optimum timing and fuel injection amount based on a control signal from the ECU 8, and 10 is necessary for spark discharge in a spark plug (not shown) of the internal combustion engine 1. It is an igniter that generates a high voltage at an optimum timing based on a control signal from the ECU 8 as well.

Next, in the ECU 8, the ignition timing θ _ig
The operation when setting is will be described with reference to the flowcharts shown in FIGS. The routine shown in FIG. 3 is performed at a predetermined rotation angle (for example, 120 ° C. for a 6-cylinder internal combustion engine).
Every A).

At step 100, the basic ignition timing ABSE is set based on the engine speed Ne obtained from the detection signal of the rotation angle sensor 5 and the intake air amount Q obtained from the detection signal of the air flow meter 4.

In step 200, the engine speed Ne as shown in FIG.
And the engine load Q / N are interpolated based on a two-dimensional map to set the maximum retardation amount AKMAX.

In step 300, the first control amount AKL is calculated according to the state of the internal combustion engine. The first controlled variable A
KL is a value that is always learned according to the internal combustion engine state,
The calculation method will be described in detail later.

In step 400, the second control amount AKCS is calculated based on the detection signal of the knock sensor 7. The method of calculating the second retard angle amount AKCS will also be described in detail later.

In step 500, steps 200-40
The sum of the first control amount AKL and the second control amount AKCS from the value obtained from 0 is restricted so as not to be larger than the maximum retardation amount AKMAX. That is, the first control amount A
When the sum of KL and the second controlled variable AKCS is larger than the maximum retardation amount AKMAX, the routine proceeds to step 600, where the first controlled variable AKL and the second controlled variable AKCS are calculated.
The maximum retardation amount AKMA
As X, the process proceeds to step 700. On the other hand, step 500
If the sum of the first control amount AKL and the second control amount AKCS is not larger than the maximum retardation amount AKMAX, the process proceeds to step 700.

In this way, the basic ignition timing AB is set by limiting the maximum retardation amount AKMAX so as not to exceed the maximum retardation amount AKMAX.
It is possible to prevent it from being set to the advance side from SE. In step 700, the ignition timing θ _ig of the internal combustion engine 1 is calculated according to the following formula (Formula 1).

[0027]

## EQU1 ## θ _ig = ABSE-AKMAX + AKL + AKCS In step 800, an ignition signal that results in the ignition timing θ _ig calculated in step 700 is output to the igniter 10, and this routine ends.

Next, step 3 in the routine of FIG.
The operation of calculating the first controlled variable AKL of 00 will be described. Here, the first control amount AKL is a value that is constantly learned according to the state of the internal combustion engine 1, and the method of setting the update amount ΔAKL is a feature of the present invention and is set using fuzzy reasoning. It

That is, a membership function is used to represent an index with an ambiguity as to whether the engine speed Ne obtained from the detection signal of the rotation angle sensor 5 is currently in a high rotation range or a low rotation range. Further, the inference rule (control law) at this time is defined as follows.

If If reliability μ _f is high and second control amount AKCS is large, then the first control amount AKL is largely updated. Here, the reliability μ _f is the engine speed Ne, the engine load Q /
It is a value set from three input conditions (membership function) of N and engine speed change amount ΔNe, and more specifically, the reliability obtained for each input condition from the two-dimensional maps of FIGS. 7A to 7C. The reliability μ _f is set by multiplying all the degrees μ _n (n = 1 to 3) as shown in Expression 2.

[0031]

(2) μ _f = μ ₁ × μ ₂ × μ ₃ That is, under the input conditions shown in FIG. 7A, when the engine speed Ne is small, the output signal of the knock sensor 7 is small, and therefore the accuracy of knocking detection is Therefore, the reliability μ ₁ becomes 0 or near 0, while the engine speed Ne
When is large, the vibration of the internal combustion engine 1 itself is large, and therefore it may be erroneously detected that knocking has occurred. Therefore, the reliability μ ₁ is set to 0 or close to 0. Therefore, at such an engine speed Ne, the update amount ΔAKL is not set large, and the first control amount AKL is not updated significantly. Further, here, for the reason described above, the engine speed N
The reliability μ ₁ is highest when e is around 2000 rpm (μ ₁
= 1) is set, and at this time, the first control amount AKL is largely updated (learned).

Further, the engine speed Ne of 0 to 2000 to 4
The magnitude of the engine speed Ne with ambiguity is expressed by forming a triangle with 2000 rpm as the apex with respect to a change of 000 rpm.

Next, in FIG. 7B, the engine load Q
Generally, when / N is small, the engine is in an internal combustion engine state in which knocking is unlikely to occur, and since it is not desired to learn the first controlled variable AKL greatly on the basis of information under such a state, the reliability μ ₂ is 0 or 0. On the other hand, on the other hand, when the engine load Q / N is large, knocking is likely to occur and the frequency of such an operating state is low. Therefore, the first control amount AKL is largely learned based on the information at this time. The reliability μ ₂ is set to 0 or near 0 because it is not desired.

Next, in FIG. 7 (c), the engine speed
The change amount ΔNe is large, that is, the internal combustion engine 1 is in a transient state.
Sometimes, the knocking detection accuracy deteriorates.
Preventing the first controlled variable AKL from being greatly learned
For reliability μ₃Becomes 0 or near 0, and vice versa
The most reliable value μ under the condition that the amount of change in the rotational speed ΔNe is small. ₃
Is set large.

Further, according to the inference rule of this embodiment, the update amount ΔAKL is set large when the second control amount AKCS is large. That is, the second controlled variable AKCS
Will be described in detail later, the value is updated depending on whether knocking has occurred. If the second control amount AKCS is set to a large value, knocking continuously occurs or knocking occurs. It is an internal combustion engine state in which it does not occur continuously, and under such a state, the ignition timing θ
_{Although the ig} can be largely updated, the second control amount AKCS is only updated by the predetermined amount α in the conventional method. However, by using the control method of the ignition timing θ _ig based on the above inference rule, the first control amount AKL is largely updated by setting the update amount ΔKCL to a large value, and the optimal ignition timing θ _ig is obtained with good responsiveness. It can be controlled.

Next, the process of actually setting the first controlled variable AKL based on the above-mentioned inference rule (step 300).
Will be described with reference to the flowchart in FIG. In step 301, the reliability μ ₁ at the current engine speed Ne is obtained from the two-dimensional map shown in FIG. 7A, and in step 302, the current engine load Q / is calculated from the two-dimensional map shown in FIG. 7B. The reliability μ ₂ at N is obtained, and at step 303, the reliability μ ₃ at the current engine speed change amount ΔNe is obtained from the two-dimensional map shown in FIG. 7C.

In step 304, the update amount ΔAKL based on the second control amount AKCS is calculated from the two-dimensional map shown in FIG.
To set. Note that, here, the reliability μ is set in the same manner as when the update amount ΔAKL is set for the engine speed Ne described above.
₄ is defined, and the second control amount AKCS uses the value obtained when the routine of FIG. 3 was executed last time.

By the way, the reliability μ ₄ is expressed by a positive region and a negative region as shown in FIG. 8, and the first control amount AKL is greatly increased in the positive direction, that is, the first control amount AKL is increased. When learning in the direction, the reliability μ ₄ is set large in the positive region. On the other hand, when learning the large first control amount AKL in the negative direction, that is, in the direction of decreasing the first control amount AKL, the reliability μ ₄ is set large in the negative region.

Therefore, by expressing the reliability μ ₄ in the positive region and the negative region in this way, the'first control amount A
The contradictory controls of “decreasing KL” and “increasing the first control amount AKL” can be expressed by one map (membership function), and the arithmetic processing can be simplified.

Then, in step 305, the overall reliability μ _f is obtained by multiplying the reliability μ _n (n = 1 to 4) obtained by the above method. Here, a specific description will be given to clarify the above-mentioned processes, for example, engine speed Ne = 1600 rpm, engine load Q / N = 1.0 l / r.
ev, the engine speed change amount ΔNe = 200 rpm / s, and the second control amount AKCS = 2 ° C., FIG.
From (c) and the two-dimensional map of FIG. 8, the reliability μ _n (n =
1-4) are μ ₁ = 0.8, μ ₂ = 1.0, μ
₃ = 0.9 and μ ₄ = 0.8.

Therefore, the overall reliability μ _f can be obtained from Equation 3.

[0042]

## EQU3 ## μ _f = μ ₁ × μ ₂ × μ ₃ × μ ₄ = 0.8 × 1.0 × 0.9 × 0.8 = 0.576 In step 306, the first value corresponding to the reliability μ _f The update amount ΔAKL of the control amount AKL is calculated from the two-dimensional map shown in FIG. That is, if μ _f = 0.576, then 2 in FIG.
From the dimensional map, the update amount ΔAKL = 0.576. Here, also in the two-dimensional map of FIG. 9, the reliability μ _f is represented by a positive region and a negative region as in the case of FIG. 8, and contradictory control is represented by one map to simplify the arithmetic processing. .

At step 307, the first control amount AKL is learned based on the update amount ΔAKL calculated at step 406 using the following equation (equation 4).

[0044]

[Formula 4] AKL = AKL + ΔAKL In step 308, the newly learned first control amount AKL
Is controlled to be 0 or less, and the first controlled variable AKL
If is less than or equal to 0, proceed to step 309, where the first
The control amount AKL of is set to 0 and the process proceeds to step 400.
By adding such processing, the maximum retardation amount AKM
It is possible to prevent the ignition timing θ _{ig from} being set on the retard side of AX. On the other hand, if the first control amount AKL is not 0 or less, the process proceeds to step 310.

In step 310, it is judged whether or not the newly learned first control amount AKL is larger than the maximum retardation amount AKMAX, and the first control amount AKL is the maximum retardation amount AKMAX.
If it is larger, the process proceeds to step 311, and the first control amount A
Set KL to the value of maximum retardation amount AKMAX and step 4
Go to 00. By adding such processing, it is possible to prevent the advance ignition timing ABSE from being set this time.

Next, step 4 in the routine of FIG.
The process of 00, that is, the operation of calculating the second control amount AKCS will be described with reference to the flowchart shown in FIG.

At step 401, based on the knocking determination result in a knocking determination routine (not shown),
If knocking has occurred, proceed to step 402,
If there is no knocking, step 40 as it is.
Go to 3. The knocking determination as to whether or not knocking has occurred in the internal combustion engine 1 is performed by a known method such as comparing a detection signal of the knock sensor 7 with a knock determination value.

In step 402, the second controlled variable AKCS
Is decreased by a predetermined amount α (for example, α = 0.5 ° C. A), and the process proceeds to step 403. In step 403, a timer (not shown) is checked and every 1 second, the process proceeds to step 404, where the second
Control amount AKCS is increased by a predetermined amount α and step 405
Proceed to.

In step 405, the second controlled variable AKCS
A limit value LAKCS for limiting the update of
Specifically, the reliability μ _n (n =
Value multiplied by all 1 to 3) μ _f0 (μ _f0 = μ ₁ × μ ₂ ×
Based on μ ₃ ), it is calculated from the two-dimensional map as shown in FIG.

At step 406, the second controlled variable AKCS
Is larger than the limit value LAKCS, and if larger, the routine proceeds to step 407, where the value of the second control amount AKCS is set to the limit value LAKCS and the routine proceeds to step 408. On the other hand, if the second control amount AKCS is not larger than the limit value LAKCS, the process directly proceeds to step 408.

At step 408, the second controlled variable AKCS
Is less than the limit value- (LAKCS),
If it is smaller, the process proceeds to step 409, where the second control amount AKC
The routine ends with the value of S being the limit value- (LAKCS), and proceeds to step 500.

Therefore, as described above, the second control amount AK
CS sets the update range to -LAK by the limit value LAKCS.
It is limited to CS <AKCS <LAKCS, and by limiting in this way, even if an erroneous determination is made in the knocking determination, the optimal ignition timing θ _ig does not deviate by more than a predetermined range.

The limit value LAKCS is the reliability μ _f0.
The limit value LAKCS is set to a large value in the operating state in which knocking is unlikely to occur and the knocking detection accuracy is high, and conversely, in the operating state in which knocking easily occurs and the knocking detection accuracy is low. Since the limit value LAKCS is set small, it is possible to set an appropriate limit value LAKCS according to the operating state.

FIG. 11 shows the ignition timing θ using the above-mentioned method.
11 is a diagram showing a setting method of _{ig. As} shown in FIG. 11, the ignition timing θ _ig is once retarded to the maximum retardation amount AKMAX, and thereafter, the first control amount AKL and the second control amount AK are set.
It is set by correcting the lead angle side with CS (Equation 1).

As described above, in this embodiment, the conventional method is used.
Ignition timing θ of the internal combustion engine 1 unlike the law _igIs the prescribed engine speed
In the high rotation range of Ne or higher, the maximum retardation amount AK is always
It is not fixed to MAX, depending on the state of the internal combustion engine.
Reliability μ_fAnd the reliability μ _fThe larger the value, the greater the ignition
Period θ_igLearn the first controlled variable AKL so that
Therefore, the optimum ignition timing θ_igIs set and high
It is possible to prevent the output from decreasing in the rotation region.

Further, when knocking does not occur continuously, when the ignition timing θ _ig is updated from the maximum retardation amount AKMAX to the advance side, the first control amount AKCS and the internal combustion engine state are fuzzy. Since it is updated by the second control amount AKL set by using inference, it is possible to control the ignition timing θ _ig to have an optimal responsiveness.

In this embodiment, the reliability μ ₄ based on the setting of the update amount ΔAKL from the update state of the second control amount AKL.
However, the reliability μ ₄ may be set according to the knocking determination result obtained from the output signal of the knock sensor 7. In this case, for example, knocking occurs continuously or continuously. The reliability μ ₄ is set based on the information that it has not been done.

Next, another method of setting the ignition timing θ _ig of the internal combustion engine 1 will be described below with reference to the drawings. That is, in the second embodiment, the value of the first controlled variable AKL is the engine speed N
The ignition timing θ _ig is stored in each region determined by e, and the ignition timing θ _ig is the first control amount AK in the region corresponding to the current engine speed Ne.
An embodiment when the present invention is applied to an ignition timing control device which is set from L and which has a configuration of learning the value of the first control amount AKL in this corresponding region will be described.

That is, by performing learning for each region in this way, it is possible to set the optimum first control amount AKL in all operating regions as compared with the case where no division is made into each region, and thus accuracy is improved. That is, the optimum ignition timing θ _ig can be set well.

By the way, the value of the first controlled variable AKL is, as described above, learned when the operating state to which this value belongs belongs. However, depending on the actual operating state, the value of the first operating state The region is always learned, but conversely, the region in the second driving state is rarely learned, which causes a difference in learning frequency. In this case, although the learning function is provided, the first control amount AKL in the region where the learning frequency is low is not always set to the optimum value and the ignition timing θ _ig cannot be set effectively. The problem arises.

Therefore, in order to solve such a problem, the update amount of the region with high learning frequency is reflected in the update amount of the region with low learning frequency, and at this time, the region with high learning frequency and the region with low learning frequency are The exponent with ambiguity such as the boundary of is expressed by a membership function.

Therefore, here, the first control amount AKL is set for each engine speed Ne, specifically, for each of the three regions (high rotation region), (medium rotation region), and (low rotation region). I will explain what you have. The inference rule is defined as shown below.

(1) If If the second control amount AKCS is large in the low rotation region, then the first control amount AKL in the low rotation region is increased. (2) If the second control amount AKCS is large in the medium rotation region If, then the first control amount AKL in the middle rotation region is increased.

(3) If If the second control amount AKCS is large in the high rotation region, then the first control amount AKL in the high rotation region is increased. Here, the low speed region, the index such as middle speed region, and the high speed region, or index, such that increasing the second control amount AKL is reliability determined in accordance with the engine speed Ne with reference to FIG. 12 mu _i ( i = a to c).

More specifically, for example, engine speed
Ne = 2000 rpm, second control amount AKCS = 2 ° C.
8 and FIG. 12, the second controlled variable AKC
Confidence that S is large μ_FourBecomes 0.8, and (low rotation
Domain) reliability μ _aIs 0.6,
Range) reliability μ_bIs 0.5, (high rotation range)
Confidence μ_cBecomes 0. Therefore, the reliability μ
_fIs (high rotation area), (medium rotation area), (low rotation area)
Each of the three areas of
The reliability μ is calculated as 8, 0.40, 0._fFrom update amount
ΔAKL is calculated for each area.

[0066]

[Mathematical formula-see original document] μ _f = μ ₄ × μ _i (i = a to c) FIG. 13 shows that the first control amount AKL is actually learned, and
4 is a flowchart showing the operation when calculating the first controlled variable AKL under the current operating condition, and describes the processing of step 300 in the routine of FIG. 3. Regarding the method for setting the ignition timing θ _ig, the other processing is the same as that shown in FIGS. 3 and 5.

[0067] Based on the step 351 FIG. 12, obtains the reliability mu _c confidence mu _a, reliability mu _b of the middle speed region and high speed region of the low rotation region using the above-described method. In step 352, the reliability μ ₄ indicating whether the second control amount AKCS is large is obtained based on FIG. 8, and in step 353, the reliability μ _f is calculated based on the above equation 4,
The update amounts ΔAKL of the three regions are set by using the method already described with reference to FIG.

At step 354, the first control amount AKL for each of the three regions is learned based on the update amount ΔAKL set at step 353. The first control amount AK
The learning of L is performed based on the above-mentioned formula 3.

In step 355, the first control amount AKL actually used for setting the present ignition timing θ _ig is multiplied by the reliability μ _i (i = a to c) of each region and the first control amount AKL. Find from the center of gravity of. That is, it is set based on the following formula (Formula 6).

[0070]

## EQU6 ## AKL = (μ _a × AKL _L + μ _b × AKL _M + μ _c × AKL _H ) / μ _k μ _k = μ _a + μ _b + μ _c where AKL _L , AKL _M , and AKL _H are step 35
It is the first control amount of the low rotation region, the medium rotation region, and the high rotation region learned in 4 respectively.

As described above, when the first control amount AKL is divided into three regions and set, the boundary portion of these regions is expressed by the membership function and the second region is set.
Control amount AKCS is large, or the first control amount AKL
By expressing an index with ambiguity, such as increasing the value, by a membership function, it is possible to effectively judge which region the current operating state corresponds to, and further, to which this operating state corresponds. It is possible to change the reflection amount for each region according to the operating state without uniformly reflecting the learning value ΔAKL reflected in the other regions in all regions.

In this embodiment, the first control amount AKL is stored for each of the three regions in accordance with the engine speed Ne, but a plurality of three or more first control amounts AKL are stored. It may be stored, and the engine speed N
The first control amount AKL may be stored for each of the regions determined by the engine load Q / N and the engine speed Ne without setting the region only with e. The boundary portion of the engine load Q / N at this time is represented by a membership function.

In this embodiment, the maximum retardation amount AKM is once set.
Although the ignition timing θ _ig is set by retarding the ignition timing to AX and then correcting the ignition timing θ _ig to the advance side by the first control amount AKL and the second control amount AKCS, the first retardation amount AKMAX is not used. The ignition timing θ _ig may be set by controlling the ignition timing θ _ig to the retard side from the basic ignition timing ABSE based on the control amount AKL of the above and the second control amount AKCS.

[0074]

As described above, in the present invention, the operating state of the internal combustion engine and the knocking determination result are expressed as a membership function by a membership function, and the learning amount is obtained by using fuzzy inference, and the learning amount is calculated. First based
By learning the control amount of, and always correcting the ignition timing based on the first control amount, the ignition timing is always set to the maximum retard side in the high rotation region, or with a constant retard amount. It is not controlled, and has an excellent effect that the optimum ignition timing of the internal combustion engine can be always set in all operating regions according to the knocking determination result and the internal combustion engine state.

Further, there is also an excellent effect that the ignition timing can be controlled to the optimum ignition timing with good responsiveness according to the state of the internal combustion engine.

[Brief description of drawings]

FIG. 1 is a diagram corresponding to a claim of the present invention.

FIG. 2 is a schematic configuration diagram showing a configuration of an embodiment of the present invention.

FIG. 3 is a flowchart for explaining the operation of the present invention.

FIG. 4 is a flowchart for explaining the operation of the present invention.

FIG. 5 is a flowchart for explaining the operation of the present invention.

FIG. 6 is a two-dimensional map of engine speed-engine load used for the explanation of the process in FIG.

FIG. 7 is a diagram expressing a membership function used to explain the processing of FIG.

FIG. 8 is a diagram expressing a membership function used to explain the processing of FIG.

FIG. 9 is a two-dimensional map for obtaining the update amount used for the explanation of the processing in FIG.

FIG. 10 is a two-dimensional map for obtaining a limit value used for the explanation of the processing in FIG.

FIG. 11 is a diagram showing a method of setting an ignition timing according to the present invention.

FIG. 12 is a diagram expressing a membership function according to another embodiment of the present invention.

FIG. 13 is a flow chart for explaining the operation of another embodiment of the present invention.

[Explanation of symbols] 1 Internal combustion engine 3 Throttle valve 4 Air flow meter 7 knock sensor 8 Electronic control unit (ECU) 10 Igniter

Claims (2)

[Claims] 1. An internal combustion engine state detecting means for detecting an operating state of an internal combustion engine, a knock determining means for determining whether knocking occurs in the internal combustion engine, and a basic of the engine according to an operating state of the internal combustion engine. A basic ignition timing calculation means for calculating an ignition timing, a detection result of the internal combustion engine state detection means and a determination result of the knock determination means are expressed as a membership function as an input condition, and a learning amount is obtained using fuzzy reasoning. A first control amount learning means for learning a first control amount for controlling the basic ignition timing based on the learned amount, and a second control amount for controlling the ignition timing based on the determination result of the knock determination means. A second control amount calculating means for calculating a control amount; and the internal combustion engine by correcting the basic ignition timing according to the first control amount and the second control amount. Ignition timing control apparatus for an internal combustion engine characterized by comprising an ignition timing setting means for setting the ignition timing. 2. The plurality of first control amounts are set for each region according to the operating state of the internal combustion engine, and are learned for each region using the fuzzy inference. 1. An ignition timing control device for an internal combustion engine according to 1.