JPS59119068A - Knocking controlling method for internal-combustion engine - Google Patents

Abstract

PURPOSE:To reduce the effect of noise, by setting a change for a learned delay to be not larger than a predetermined value. CONSTITUTION:In step 118, it is judged whether or not a corrected delay thetak is not smaller than 2 deg.CA. When the corrected delay is smaller than 2 deg.CA, a learned value alpha is made to be -0.04 deg.CA in step 160, and then step 162 is entered. When thetak is not smaller than 2 deg.CA in step 118, it is judged in step 119 whether or not thetak is not smaller than 4 deg.CA. When thetak is not smaller than 4 deg.CA, the learned value alpha is made to be 0.04 deg.CA in step 161. Accordingly, the corrected delay is set in a predetermined range, for example, 2 deg.CA<thetak<4 deg.CA, so that the effect of noise is reduced.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a knocking control method for an internal combustion engine, and more particularly, to a correction retardation amount for performing retarding at a relatively fast speed depending on the presence or absence of knocking, and a method for controlling knocking at a relatively slow speed depending on the presence or absence of knocking. The present invention relates to a method for controlling knocking by correcting a basic ignition advance angle by retarding the ignition angle and using a learning retard amount changed by learning control.

The conventional knocking control method using learning control is based on the engine speed N1, the ratio Q of the intake air amount Q and the engine speed N,
/N or the load determined by the intake pipe negative pressure, the basic ignition advance angle θBA8K is stored in the read-only memory (ROM) of the micro comviator in Mazda form, and is actually calculated based on the following formula (1). The ignition advance angle θ1g for controlling the igniter is calculated, and knocking control is performed using this ignition advance angle.

θig= oBABK - (θKG+θK)...
(1) Knocking occurs at θ, a learning retard amount determined by the engine speed and load and changed by learning control in order to bring the knocking level to a predetermined level. This is the correction retard amount that retards the ignition timing and advances the ignition timing when knocking no longer occurs.

Here, the corrected retard amount θ is determined as follows.

First, engine vibration is detected using a knocking sensor composed of a microphone, etc., and a predetermined times x - b (where K is a proportionality constant) of the average value (background) of engine vibration b is calculated from the peak value a of engine vibration. This peak value a is compared with the value of -t). When the peak value exceeds the value of a(-1), it is determined that knocking has occurred, and the ignition timing is adjusted to a predetermined crank angle (for example, 0.4℃A) per knocking occurrence, as shown in the following equation (2). The correction retard amount θK is changed so as to cause a delay.

To θ←θK + 0.4℃A ・・・・・・・・・
(2) Also, when the peak value a f)'h K'' b is lower than the value, it is determined that knocking has not occurred, and the first timer is used for a predetermined period of time (131, for example, 4
B m5ec), and when the predetermined time has elapsed, the correction retard amount θK is set so that the ignition timing advances by a predetermined crank angle (for example, 0.08°C A) as shown in the following equation (3). change.

To θ←θK -0.08℃A ・・・・・・・・・
...(3) Also, the learning retardation amount θ according to the engine conditions
KG is calculated as follows. First, as shown in Fig. 1, addresses θ to 23 at which learning retardation amounts are stored in correspondence with engine speed N and load Q, /N are prepared in the random access memory (RAM) of the microcomputer and the learning is performed. Create a map. The engine speed N and intake air amount Q are taken in, and the 4 points RA surrounding the point (N%Q'/N) indicating the current engine condition on the learning map
Find the address of M. Now, as shown in Figure 2, the RAM address surrounding the point indicating the current engine condition is n (n=Q, l
...16), n+1, n+6, n+7,
Learning retardation amount θKGn at address n, learning retardation amount θKG in+1) at address n+1, learning retardation amount θKG at address n+6
ln+61, the learning delay amount θxa(n+73
It is assumed that each of them is memorized. Then, the difference in engine speed between addresses x1 is the difference in load between addresses, and the difference in engine speed between address Y1 and the point indicating the current engine condition is the difference in engine speed between address x1 and the point indicating the current engine condition. If the difference in load between the two points is y, then the learning retardation amount θKG of the point indicating the current engine condition can be determined by the two-dimensional interpolation method shown in equations (4) to (6) below.

(5) The learning control of the learning Mazda described above is performed as follows. First, a second timer that determines the learning control time depending on the current engine conditions and a third timer that determines the learning control time regardless of the engine conditions are prepared. A second timer sets a predetermined time (for example, 48 m (6)
), when it is detected that the correction retard amount θK
is changed and exceeds a predetermined crank angle (for example, 4°C A), and when the corrected retardation amount θK exceeds the predetermined crank angle, the point indicating the current engine condition explained above is surrounded. The learning retardation amount of the four points on the learning map is added to a predetermined crank angle (for example, 0.04° C.A.).

As a result, the learning retard amount is learning-controlled so that the ignition timing is delayed. On the other hand, a third timer is used for a predetermined period of time (for example, 1
6) When it is detected that the ignition timing has elapsed, K subtracts a predetermined crank angle (for example, 0.01" CA) from the learning retard amount of all addresses on the learning map, regardless of the presence or absence of knocking, and sets the ignition timing. The learning delay amount is controlled in a learning manner so that the learning progresses.

[,Using the corrected retard amount θ changed as described above and the learning retard amount θKG obtained by two-dimensional interpolation from the learning Mazda subjected to learning control, the basic calculation is performed based on the formula +11 above. Knocking is controlled by correcting the ignition advance angle θBA8E.

However, in such conventional methods, the learning retard amount corresponding to the current engine condition is calculated from the learning map by two-dimensional interpolation, so when rewriting the learning map value by learning control, momentary power outages, etc. When noise due to is input,
There may be cases where the learning map values are not rewritten accurately.
For example, as shown in FIG. 3, the learning delay amount at address n+1 may become abnormally small. In such a case, if you calculate the learning retard amount corresponding to the current engine conditions from the learning map that could not be rewritten accurately, the calculated learning retard needle will become abnormally small, indicating that the ignition timing has advanced too much. There was a problem with frequent knocking.

The present invention has been made to solve the above-mentioned problems, and it is an object of the present invention to provide a knocking control method that prevents frequent knocking even if the learning map is not rewritten accurately.

In order to achieve the above object, the present invention provides a knocking control method that performs knocking control by subtracting the sum of a corrected retard amount and a learned retard amount calculated based on a learning map from a basic ignition advance angle. A value at which the amount of change from the previously calculated learning retardation amount is less than or equal to a predetermined value is used as the value of the learning retardation amount, and knocking is controlled by the learning retardation amount whose amount of change does not exceed the predetermined value. It is something.

According to the configuration of the present invention, even if the learned retard amount that has been subjected to learning control is not stored accurately, the knocking control is performed using a value that gradually decreases the learned retard amount, so that the ignition timing is rapidly advanced. A unique effect is obtained in that the occurrence of knocking is reduced.

Here, it is preferable to calculate the learning retardation amount at every predetermined crank angle or every predetermined time. Moreover, as shown in FIG. 4, the basic ignition advance angle θBABK, that is, M B T
(Minimum 8park Advance Fo
r Be5tT Orq Lt changes as shown by curve CI depending on the engine speed, and the ignition timing at which slight knocking occurs when knocking is difficult to occur, such as when the air is humid, changes as shown in curve C1, and The ignition timing at which slight knocking occurs when knocking is likely to occur, such as when the air is dry, is as shown by curve c3, and the ignition timing at which slight knocking occurs varies depending on the engine speed and environmental conditions. Therefore, in the above configuration of the present invention,
In order to always perform the same knocking control even if the engine speed and environmental conditions change, it is preferable to perform learning control on the learning retardation amount so that the corrected retardation amount falls within a predetermined range.

Next, FIG. 5 shows an example of an engine to which the present invention is applied. As shown in the figure, this engine is equipped with an air flow meter 2 as an intake air amount sensor provided downstream of a factory cleaner (not shown). The air flow meter 2 includes two compensation motion grates rotatably provided in the damping chamber, and a potentiometer 2B that detects the opening degree of the compensation gray 2A. Therefore, the intake air iQ is detected as a voltage output from the potentiometer 2B. Further, an intake air temperature sensor 4 is provided near the air flow meter 2 to detect the temperature of intake air.

A throttle valve 6 is arranged downstream of the air flow meter 2, and a surge tank 8 is arranged downstream of the throttle valve 6.
is provided. An intake manifold 10 is connected to the surge tank 8, and a fuel injection valve 12 is disposed protruding into the intake manifold 10.The intake manifold IO is connected to the combustion chamber 14AGC of the engine body 14. The combustion chamber 14A of the engine is connected to a catalytic converter (not shown) filled with a three-way catalyst via an exhaust manifold 16. A knocking sensor!8 is provided to detect the vibration of the engine.

Note that 20 is an ignition glove, 22 is a 0□ sensor for controlling the air-fuel mixture to near the stoichiometric air-fuel ratio, and 24 is a cooling water temperature sensor for detecting the engine cooling water temperature.

The spark plug 2o of the engine body 14 is connected to a distributor 26, and the distributor 26 is connected to an igniter 28. This distributor 2
6 is provided with a cylinder discrimination sensor 30 and an engine rotation angle sensor 32, which are composed of a pickup and a signal rotor fixed to a distributor shaft.

This cylinder discrimination sensor 3o outputs a cylinder discrimination signal to an electronic control circuit 34 made up of a microcomputer or the like, for example, every 720 degrees of crank angle, and this engine rotation angle sensor 32 outputs a cylinder discrimination signal, for example, every 30 degrees of crank angle. An angular reference position signal is output to the electronic control circuit 33.

The electronic control circuit 34, as shown in FIG.
Access memory (RAM) 36, read-only memory (ROM) 38, and central processing unit (CPU)
) 40, a clock (cLocK) 41, a wJl output port 42, a second input/output port 44, a first output port 46, and a second output port 48. , ROM38, CPU40, CLO
CK41, first input/output capo) 42, a20 input/output boat 44, first output boat 46, and second output boat 48 are connected by a bus 50. The ml input/output port 42 includes a buffer (not shown), a multiplexer 54, and an analog-to-digital (A/D) converter 5.
6, air flow meter 2, cooling water temperature sensor 24
and an intake air temperature sensor 4, etc. are connected. This multiplexer 54 and A/D converter 56 are controlled by a signal output from the first input/output port 42.

Second input/output boat 44 Niha1. <0.0 through the sofa (not shown) and the comparator 62. cylinder discrimination sensor 3o is connected to cylinder discrimination sensor 3o via waveform shaping circuit 64.
and an engine rotation angle sensor 32 are connected.

Further, the knocking sensor 18 is connected to the second input/output port 44 via a bandpass filter 60, a peak hold circuit 61, a channel switching circuit 66, and an A/D converter 68. This bandpass filter is connected to a channel switching circuit 61) via an integrating circuit 63. This channel switching circuit 66 trap is a control output from the second output converter 44 for inputting either the output of the peak hold circuit 61 or the output of the integrating circuit 63 to the A/Dv converter 68. A signal is input to the peak hold circuit 61, and a reset signal is input from the second input/output port 44 to the peak hold circuit 61.

Further, the first output boat 46 is connected to the ignite 28 via the drive circuit 7o, and the second output boat 48 is connected to the fuel injection port 4112 via the drive circuit 72.

The ROM 3B of the electronic control circuit 34 stores in advance a map of the basic ignition advance angle θBAIIK expressed by the engine speed and intake air amount, the basic fuel injection amount, etc.
The basic ignition advance angle and basic fuel injection amount are read based on the signals from the air flow meter 2 and the engine rotation angle sensor 32, and various signals including the signals from the cooling water temperature sensor 24 and the intake air temperature sensor 4 are used to read out the basic ignition advance angle and basic fuel injection amount. 1
A corrected ignition advance angle and a corrected fuel injection amount are added to the stored main ignition advance angle and basic fuel injection amount, and the igniter 28 and fuel injection valve 12 are controlled. The air-fuel ratio signal output from the sensor 22 is used for air-fuel ratio control to control the air-fuel ratio of the air-fuel mixture to near the stoichiometric air-fuel ratio. Further, the learning Mazda shown in FIG. 1 is stored in advance in the RAM 36 of the electronic control circuit 34.

Next, an embodiment in which the present invention is applied to the engine as described above will be described in detail. In addition, in explaining the present invention in detail, the main routines of fuel injection control, air-fuel ratio control, ignition timing control, etc. are the same as conventional ones, so the explanation will be omitted, and only the knocking control routine related to the present invention will be explained. Figure 7 shows an interrupt routine every 30°C when the present invention is implemented using a computer. First, in step 81, the engine rotation speed N is calculated from the rotation time based on the signal from the engine rotation angle sensor 32, and in step 82, the number of interruptions after the cylinder discrimination signal is input from the cylinder discrimination sensor 30 is counted. Sets a flag indicating the current crank angle. Next, step 8
3, the flag set in step 82 is at top dead center (T
DC) flag. If the current position is not the top dead center, the process proceeds to step 88, and if the current position is the top dead center, it is determined in step 84 whether or not the knock gate is closed. When the knock gate is open, the knock gate is closed in step 85, and when the knock gate is closed, the channel switching circuit 66 is switched in step 86, and the engine vibration signal output from the knock sensor 18 is passed through the band pass filter 60. , through the integrating circuit 63 and channel switching circuit 66.
input to the D converter 68 and begin A/D conversion of the average value of engine vibration, ie, the background level.

Subsequently, in step 87, the knock gate is turned to the closing side t.
1, that is, calculate the next time to close the knock gate and set the time coincidence interrupt A.

Next, in step 88, it is determined based on the flag set in step 82 whether the crank angle has reached 90 degrees ABTDC (before top dead center). Crank angle is 90 degrees ABTD
If it is not C, proceed to step 91 and set the temperature to 90℃ABTDC.
In this case, in step 89, the corrected retard amount θ is updated and the ignition timing is calculated (details of this will be explained below). In step 90, the time to turn on the igniter 28 is determined based on the ignition timing calculated in step 89 and the current time, and a time coincidence interrupt B is set, and an igniter-on flag is set. Then, in step 91~1, the crank angle is 60℃ AB.
Determine whether TDC has been reached, and if it is not 60'CABTDC, return to the main routine and return to 60'CABTDC.
If it is TDC, the igniter off time is calculated in step 92, a time coincidence interrupt B is set, and the igniter on flag set in step 90 is removed.

Next, the time coincidence interrupt A shown in FIG. 8 will be explained.

This interrupt routine is for finding the peak value of engine vibration, and when the time set in step 87 in FIG. The peak hold value is input to the A/D converter 68 via the circuit 66 to start A/D conversion of the peak hold value, and the process returns to the main routine.

FIG. 9 shows the routine for time coincidence interrupt B, and the interrupt is performed when the time set in step 9o and step 92 of FIG. 7 comes. In step 94, it is determined whether the igniter on flag is set, that is, whether this flag is 1. If the flag is set, the igniter is turned on in step 96, and if the flag is off, the igniter is turned on in step 95. Turn off and return to the main routine.

FIG. 1θ shows an A/D conversion completion interrupt routine, and this interrupt is performed when the A/D conversion of the background level and the A/D conversion of the peak hold value are completed. First, in step 97, it is determined whether the knock gate is currently open. When the knock gate is closed, the seventh
The A/D conversion value converted in step 86 in the figure is stored in RAM 3.
6 memory and set it as background level b,
In step 99, the knock gate is opened and the process returns to the main routine. On the other hand, when the knock gate is open, the A/D converted in step 93 of FIG.
The converted value is stored in the memory of the RAM 36 as the peak value a, and in step 101 the knock gate is closed and the process returns to the main routine.

FIG. 11 shows an interrupt routine that is performed every predetermined time (for example, 4 m5ec) for counting the time when no knocking occurs and the time for learning control. First, in step 102, a counter MB is used to calculate the time when knocking does not occur.
1 is incremented by 1, and in step tOa, the count value of counter MB20, which calculates the learning control time, is incremented by 1. In the next step 104, the count value of the counter TIMEI is 12 (48 msec).
Determine whether the following is true. Count value is 12
If the count value exceeds 12, the count value of the counter MB1 is set to 12 in step 105.
In the following cases, go to step 106 [Counter TIM
It is determined whether the E20 count value is 12 or less. Here, if the count value exceeds 12, the count value of the counter TlMB2 is set to 12 in step 107 and the process returns to the main routine, and if the count value is 12 or less, the process returns to the main routine.

Next, the detailed routine of step 89 in FIG. 7 will be explained based on FIG. 12. When it is determined in step 88 of FIG. 7 that the crank angle has reached 90 degrees ABTDC, the load Q/N is changed to 0.6 (1/rev
), that is, whether or not it is in the knocking control region. When the load is less than 0.6 [)/rev], that is, in the knocking non-control region, in step 123, the basic ignition advance angle θBA8g is set to the ignition advance angle θ1g,
When in the knocking control region, in step 108,
the peak value a stored in step 100 of FIG.
A value obtained by multiplying the background level b stored in step 98 of FIG. 1θ by a constant K is compared with −b. When the peak value a exceeds the value b, it is determined that knocking has occurred, and in step 110, the corrected retardation amount θK is increased by a predetermined angle (for example, 0.4° C.A.), and in step 112, knocking has occurred. Clear the count value of the counter TIMEI, which counts the time when no operation occurs.

On the other hand, when the peak value a is less than or equal to the value x-b, it is determined that knocking does not occur, and in step 109 it is determined whether the count value of the color/return T-engine Mll is greater than or equal to a predetermined value (12). , when the count value is greater than or equal to a predetermined value, it means that a state in which no knocking occurs has continued for a predetermined period of time, so in step 111, the correction retard amount θK is decreased by a predetermined angle (for example, 0.08° C.A.). After that, in step 112, the counter TIMF11 is cleared. Further, if the count value is less than the predetermined value in step 109, the process proceeds to step 113. In step 113, the corrected retard amount θ obtained as described above is
The basic ignition advance angle θBA8K is corrected as shown in equation (1) above using the learning retardation amount θKG obtained by two-dimensional interpolation from the learning map, and the ignition advance angle 01g that actually controls the igniter is calculated. do.

Next, a routine for determining the learning retardation amount θKG corresponding to the current engine condition from the learning pine 1 and performing learning control will be described. FIG. 13 shows this routine from the middle of the main routine.

First, in step 124, the load Q/N is 0.6
(1/rev) or more, that is, whether or not it is in the knocking control region. If it is not in the knocking control area, proceed directly to the main routine, and if it is in the knocking control area, enter the RAM addresses of the four points surrounding the point indicating the current engine condition determined by engine speed N and load Q/N in Stella l114. Ask on learning Mazda. Next, in step ttS, a two-dimensional interpolation method (two-dimensional interpolation method (The routine will be explained later) calculates the learning retardation amount θKG at the point indicating the current engine condition, and stores the calculated value in a predetermined location in the RAM. Step 116
Then, it is determined whether the counter 'I'IME20 count value counted in step 103 of FIG. 11 for determining the learning control time is equal to or greater than a predetermined value (for example, 12). If the count value is less than a predetermined value, the process returns to the main routine, and if the count value is greater than or equal to the predetermined value, the count value of the counter MFlt2 is cleared in step 117, and then steps 110 and 11 in FIG.
Step 118
Judge by.

If it is determined in step 118 that the corrected retardation amount θK is less than the first predetermined crank angle, then in step 121 the corrected retardation amount θK is stored at four points on the learning map surrounding the points indicating the current engine condition.・From each learning delay amount,
Learning control is performed to subtract a predetermined crank angle (for example, 0.04°C A), and the process returns to the main routine. As a result, when the corrected retard amount θK is less than the first predetermined crank angle, learning control is performed so that the learning retard amount of the learning map becomes smaller, and the ignition timing is controlled to advance by the learned retard amount. On the other hand, if it is determined in step 118 that the corrected retardation amount θK is equal to or greater than the first predetermined crank angle,
In step 119, it is determined whether the corrected retardation amount θK is less than a second predetermined crank angle (for example, 4° C.A.) that is larger than the first predetermined crank angle. Step 1t9
If it is determined that the corrected retard amount θK is less than the 11 predetermined crank angle, that is, if the corrected retard amount θK satisfies the following conditions, return to the main routine without learning control.

As a result, when the corrected retard amount θK takes a value within a predetermined range, learning control is performed, and the ignition timing is not changed depending on the learned retard amount. Note that even when the corrected retard amount takes a value within a predetermined range, learning control may be performed as necessary. If it is determined in step 119 that the corrected retard amount θK is greater than or equal to the second predetermined crank angle, step 12
At 0, learning control is performed to add a predetermined crank angle (for example, 0.04°C Return to routine.

As a result, when the corrected retardation amount θK is equal to or greater than the second predetermined crank angle, learning control is performed such that the learning retardation amount of the learning Mazda becomes thicker, and the ignition timing is controlled to be delayed by the learned retardation amount. Ru.

By performing the learning control as described above, the learning retard amount of the learning Mazda is changed so that the corrected retard amount falls within a predetermined range.

The learning routine shown in FIG. 13 will be explained in detail below.

FIG. 14 shows a detailed route of the two-dimensional interpolation method in step 1150 of FIG. 13. In this two-dimensional interpolation routine, the Mazda shown in Figure 1 is used as the learning Mazda, and the four RAM addresses representing the current engine conditions are set to n, n+1, n+6, and n+7 as shown in Figure 2. do. First, in step 130, the current load Q
/ N is the upper limit of the load on the learning Mazda i.e. 1-2
It is determined whether it is less than or equal to [ij/rev]. If the load exceeds 1.2 (7/rev), 1.2 is stored in register n in step 131, and the load exceeds 1.2.
If it is less than (ll/rev), step 13
4, store the current load Q/N value in register n. In step 135, the current engine speed N is determined to be the upper limit of the engine speed on the learning Mazda, that is, 6000 (
rpm) or less. If the engine speed exceeds 6ooo (rpm:), 6000 is stored in register m in step 136, and if the engine speed is below 60.00 (rl) m), the current value is stored in step 137. Store the value of engine rotation speed N in register m. In step 138, the value of register n is set to the lower limit of the load on the learning map, that is, 0.6 (1/rev
) or more, and if the value of register n is 0.
If the value is less than 6, the value of register n is set to 0.6 in step 139, and if the value of register n is 0.6 or more, the process proceeds to step 140. And step 14
0, it is determined whether the value of register m is equal to or greater than the lower limit value of engine rotation speed on the learning map, that is, too (rpm), and if the value of register m is less than 1000, the value of register m is changed in step 141. 1000
If the value of the register m is 1000 or more, the process proceeds to the next step 142. As a result of the above, when the current engine speed N and load Q/N are values on the learning map, those values are stored in registers m and n, respectively, and the current engine speed N and load Q,/N are learned. When the upper and lower limits of Mazda are exceeded, the upper and lower limits are stored in registers m and n, respectively.

Steps 142 to Stella 1149 are routines for selecting four points on the learning map.

First, in step 142, the value obtained by subtracting the load value 0.6 (7/rev) at address 0 from the value of register n is stored in register n. Next, in step 143, the value of register n is set to the load scale interval of 0, 2 (7
3/rev) and store the integer part of the quotient in register n.
and store the remainder of the quotient in register y. The value of this register y is equal to the load value y from address n in FIG. 2 to the point representing the current engine condition. Also,
The integer part of the quotient stored in register n is the column of addresses closest to the point indicating the current engine condition and below the point indicating the current engine condition (horizontal sequence of addresses, e.g. 0 to 5).
The first column is the address sequence).

Then, in step 144, the value of register y is further divided by 0-2 (//rev). Therefore, the value corresponding to y/Y in the above-mentioned equation (6) is finally stored in register y.

In step 145, the value obtained by subtracting the engine rotational speed value 101000 (rp) at address 0 from the value of register m is stored in register m, as described above.

Next, in step 146, the value in register m is divided by looo (rpm), which is the scale interval of the engine rotational speed, and the integer part of the quotient is stored in register m, and the remainder of the quotient is stored in register X.

The value of this register X is equal to the engine speed value X from address n in FIG. 2 to the point indicating the current engine condition. In addition, the integer part of the quotient stored in register m is the row of addresses closest to the point indicating the current engine condition (and below the point indicating the current engine condition) (vertical arrangement of addresses, e.g. 0, 6 .12.18 is the first line).Then, in step 147, the value of register X is further divided by 1000 [rpm].

Therefore, in the end, register
) The value corresponding to x/X in the equation is stored.

Next, in step 148, the value of register n is multiplied by six and stored in register n, and in the next step 149, the value of register n and the value of register m are added and stored in register n. As a result, the addresses of the lower left corner of the four points surrounding the current engine condition, that is, address number n in FIG. 2, are determined and stored in register n.

In step 150, the learning retard amount θKGn stored at address n on the learning Mazda is read out from the register A.
The learning retardation amount θ stored at address n+1
Read KG (n-1-1) and store it in register B trap,
Learning retardation amount θxc(n+6
) is read out and stored in register C, and the learning retard amount θxG(n+7) stored at address n+7 is read out and stored in register D. In step tSt, the value of register B is subtracted from the value of register A, the value of register X is multiplied, and the value of register A is added to the value, and the result is stored in register E. Also, step 15
2, the value of register D is subtracted from the value of register C, the value of register X is multiplied, and the value of register C is added to this value, and the result is stored in register F14.

Finally, in step 153, the value of register F is subtracted from the value of register y, multiplied by the value of register y, and the value of register E is added to that value to learn a point indicating the current engine condition. Let the retard amount be θKG.

Next, FIG. 15 shows a detailed routine of the steps 118 to 121 shown in FIG. 13. Note that the register n in FIG. 15 is the register n of the two-dimensional interpolation routine in FIG. 14. First, in step ttS, it is determined whether the corrected retard amount θK is 2° C.A or more in the same manner as described above. If the corrected retardation amount θK is less than 2°C A, the learned value α is set to 10, o4°C A in step 160, and the process proceeds to step 162. If the corrected retard amount θK is 2°C or more, in step 119 the corrected retard amount θK is set to 4.
Determine whether the temperature is ℃8 or higher. The correction retard amount θK is 4
If the corrected retard angle t0K is less than 4°C, the process returns to the main routine, and if the corrected retard angle t0K is 4°C or more, the learned value α is set to 0.o4'CA in step 161, and step 162
Proceed to. In step 162, it is determined whether the load Q/N is greater than or equal to the lower limit value of the learning map, 0.6 [71/rev]. If the load is 0.6 or more, step 16
At 6, the engine speed N is the learning Mazda lower limit xo
'oo (rpm) or more is determined, and the load is 0.
．． If it is less than 6, it is determined in step 163 whether the engine speed N is 1000 [rpm] or less.

In Stella 1166, the engine speed N is xooo(
If it is determined that rpm: is less than 1, step 167
The previously learned learning delay angle θK at address n in
A learning control is performed in which a learning value α is added to GH, and in step 168 a learning control is performed in which a learning value α is added to the previously learned learning retard amount θKG (n+1) at address n+. will be returned to.

Here, as mentioned above, the current engine speed N and load Q,/N are Q/N22.6 (71/re
v ) and N<1ooo (rpm), the value of register n indicating the RAM address can take 0.6.12.18 in the first row, indicating the current engine condition. Step 1 if the point is in the above region
At step 67, the learning retardation amount at address 0.6.12.18 is learning controlled, and at step 168, the learning retardation amount at address 1.7.13.19 is learning controlled.

In step 166, the engine speed N IJ''-r
If it is determined that it is equal to or more than ooo (rpm), then in step 169, learning control is performed in which the learned value α is multiplied by the previously learned learning retard amount θKG at address n. The point indicating the current engine condition is Q/N≧0.6 [
: l/rav : If it exists in the area where l and N≧too(rpm), register n indicating the address of RAM.
Since the value of can be 0 to 23, all addresses are subject to learning control in step 169. In the next step 170, it is determined whether the value of register n is 23 or not. If the value of register n is not 23, step 1
71. In steps 173 and 174, whether the value of register n is not 17 or not, and whether it is not Il, respectively;
Determine whether it is not 5 or not. The value of this register n is 5.1
1.17.23 represents the address of the 6th line. When the value of register n is 23, the process directly returns to the main routine. The learning retardation amount at address 23 at this time is subjected to learning control in step 169. When the value of register n is 17.1115, step 172
In step 178, the value of register n is decreased by 1.
In this step, learning control is performed to add the learning value α to the learning retard amount θKG (n+7) at address n+7, and the process returns to the main routine. Therefore, the value of register n is 17.11
, 5, the learning retard amounts for addresses 17.11.5 and 23.17.11 in step 178 are respectively learning controlled. That will happen.

When the value of the register n is neither 23.17 nor Ill5, learning control is performed in step 175 in which the learning value α is added to the learning delay amount θKG(n+1) of the address n+. That is, if the RAM address obtained in step 149 of the two-dimensional interpolation routine is not on the sixth row, in step 169 and step l'75, the RAM address obtained in the two-dimensional interpolation routine and the address to the right of this address are The address is controlled by learning. In step 176, it is determined whether the value of register n is less than 17;
If the value of register n is 17 or more, the process returns to the main routine. That is, when the RAM address is 18 to 22 in the fourth column, learning control is performed on the address stored in register n and the address to the right of it in steps 169 and 175. On the other hand, if the value of register n is less than 17, that is, if there are four addresses surrounding the point indicating the current engine condition, step 1
In step 77, learning control is performed to add the learning value α to the learning delay amount θxc(n−+−s) at address n+6, and step 17
At step 8, learning control is performed to add the learning value α to the learning retard amount θKG(n+7) at address n+7. As a result, if there are four addresses surrounding the point indicating the current engine condition, step 169, step 175, step 17
7 and Stella 1178, the learning retard amounts of the four addresses mentioned above are controlled by learning.

If the load is determined to be less than 0.6 (71/rev) in step 162, it is determined in step 163 whether the engine speed is below looo (rpm), and if the engine speed is below sooo (rpm) Stellag 1 for
At step 65, 7 is subtracted from the value of register n, and at step 178, learning control is performed. The point indicating the current engine condition is Q
/ N (g, 6 (l/rev) and N≦1100
0 (rp), the value of register n is 0, so in step 178 in this case, the learning retard amount at address 0 is controlled by learning.On the other hand, when the engine speed N is If it exceeds +000 (rpm), it is determined in step 64 whether the value of register n is 5 or not. If it is not 5, 8 is subtracted from the value of register n in step 179, and 6 is subtracted from the value of register n, and step 177 and In step 178, the learning delay amount at addresses n+6 and n+7 is controlled by learning.Also, the value of register n is 5.
If so, 7 is subtracted from the value of register n in step 180, and learning control is performed in step 178. The point indicating the current engine condition is Q / N (g, e (l
/rev) and N>11000(rp:), the value of register n can take the value of the address in the first column. Therefore, when the value of register n is 5, the value of address 5 is set in step 17B. When the learning delay amount is under learning control and the value of register n is 0, 1, 2, 3, 4, step 177
In step 178, the learning delay amount of the address of the value of register n and the address to the right of it is controlled by learning.

Learning retardation amount θK in the learning routine shown in FIG. 15 above
The following table summarizes the relationship between the conditions for the corrected retard amount θ when G is updated by learning control and the increase/decrease in the learned retard amount θKG.

In addition, FIG. 16 shows the changes in the corrected retard amount θ, the learned retard amount θKG, and the ignition timing θig over time. As can be understood from the figure, when the corrected retard amount θK is within a predetermined range, the learned retard amount θKG is constant, and the corrected retard amount θK
When θK exceeds a predetermined range, the learned retard amount θKG increases, and when the corrected retard amount θK falls below a predetermined range, it decreases.

Furthermore, FIG. 17 shows variations in ignition timing corresponding to engine speed. In FIG. 16, curves 01 to C3 are the same as those in FIG. 3; It is understood that the corrected retardation amount θK is always constant whether knocking is likely to occur or knocking is unlikely to occur.

Next, with reference to FIG. 18, a routine for determining whether or not the calculated learning retardation amount has changed by a predetermined value will be described starting from the middle of the main routine. First, in step 182, it is determined whether or not the vehicle is in the knocking control region, similar to step 124 described above. If it is not in the knocking control region, the process proceeds to the main routine, and if it is in the knocking control region, it is determined in step 183 whether a predetermined crank angle (for example, 120° C.A.) has elapsed since the last learning retardation amount was calculated.

If the predetermined crank angle has not elapsed, the process proceeds to step 189, and if the predetermined crank angle has elapsed, then in step 185, four points surrounding the current engine condition are found on the learning Mazda in the same way as described above. A new learning retardation amount θKGi is calculated using the two-dimensional interpolation method from the data of the four points in the RAM, and is stored in the RAM. In the next step "186," a predetermined value (for example, 0.3° CA) is subtracted from the previously calculated learning retard amount θKG, that is, the learning retard amount θKG found in step "115" in FIG. 13, and that value is learned. It is stored as the retard amount θKG. Continuing (in step 187, the newly calculated learning retardation amount θK
It is determined whether Gi is greater than or equal to the learned retard amount θKG from which a predetermined value has been subtracted. θKG) When the condition of θKGi is satisfied, the process proceeds to step 189, and when the condition of θKG≦θKGi is satisfied, the new learning retard amount θ is set in step 188.
Store KGi as learning retardation amount θKG 1/, Step 1
Proceed to 89. As a result, if the learning retard amount changes by more than a predetermined value between when the learning retard amount was calculated last time and when the learning retard amount is calculated this time, the learning retard amount that was calculated last time changes. A value subtracted by a predetermined value is stored, and when the amount of change in the learning retardation amount is less than the predetermined value, a newly calculated learning retardation amount is stored. Then, knocking is controlled by these learning delay angle amounts. FIG. 19 is a diagram showing how the learning retardation amount is changed as described above. The broken line in the figure shows the calculated learning retardation amount when the learning map is not rewritten accurately, and since the calculated learning retardation amount is rapidly decreasing, knocking will occur frequently. On the other hand, in the case of this embodiment shown by the solid line, if the learning Mazda is not rewritten correctly, the calculated learning delay angle amount is subtracted by a predetermined value and is stored in the RAM. The amount of angle gradually decreases to prevent frequent knocking.

The next steps 189 to 194 are step 1 in FIG.
Similarly to 16 to 121, learning control of learning values stored in the learning Mazda is performed.

As explained above, according to the embodiments of the present invention, even if the engine speed or environmental conditions change, the corrected retardation amount remains within a predetermined range.
For example, since the learning retard amount is controlled to advance or retard so that 2℃A≦θ and ≦4℃A, it takes a relatively long time (16 seconds) to perform learning control to advance the angle. The learning control speed on the advance side can be made as fast as the learning control speed on the retard side, so the ignition timing will not be too late, and the once learned value will be learned next time. Since it is held until it reaches 100 g, the knocking force is reduced (and the conventional third counter program can be omitted, making it easier to use 10 g).

[Brief explanation of drawings]

Figure 1 is an explanatory diagram showing the learning map, Figure 2 is an explanatory diagram showing a point indicating the current engine condition and the four points surrounding this point, and Figure 3 is an explanatory diagram showing the learning map. 4 is a diagram showing the relationship between engine speed and ignition timing, FIG. 5 is a schematic diagram showing an engine to which the present invention is applied, and FIG. 6 is an electronic control circuit shown in FIG. 5. Figure 7 is a flowchart of the interrupt routine every 30°C, Figure 8 is a block diagram showing the interrupt routine.
Figure 9 is a flowchart of time coincidence interrupt A, Figure 9 is a flowchart of time coincidence interrupt B, GTO diagram is a flowchart of the A/D completion side input routine, Figure 11 is a flowchart showing the interrupt routine every 4 m5ec, and Figure 12 is a flowchart showing the interrupt routine every 4 m5ec. 13 is a flowchart of a routine for updating the correction retardation amount, and FIG. 13 is a flowchart of a learning control routine.
Fig. 14 is a flowchart of the two-dimensional interpolation routine, Fig. 15 is a flowchart showing details of the learning routine, Fig. Figure 17 shows the relationship between the engine speed and ignition timing, the relationship between the corrected retardation amount and the learning retardation amount, similar to that shown in Figure 3.
FIG. 18 is a flowchart showing a routine for determining whether the calculated learned value has changed by a predetermined value, and FIG. 19 is an edge diagram showing changes in the retard amount. 2... Air flow meter, 12... Fuel injection valve, 1
8...Knocking sensor, 32...Engine rotation angle sensor, 34...Electronic control circuit. Agent Tatsuyuki Unuma (and 2 others) Figure 1 Figure 2 Figure 3 N-Seki Figure 7 Figure 8 Figure 9 Figure 10 Figure 12 Figure 13 Figure 16 @ Figure 17 Figure 19 F34 listening

Claims (4)

[Claims] (1) Based on the learning retardation amount to bring the non-kissing level to a predetermined level, which is stored in advance in multiple numbers corresponding to the engine speed and load and is changed by learning control at predetermined time intervals, the current engine The learning retard amount corresponding to the engine speed and the current load is calculated, and the amount of change from the previously calculated learning retard amount is less than or equal to a predetermined value. Knocking is eliminated by subtracting the sum of the above-mentioned water-filled value and a corrected retard amount that retards the ignition timing when knocking occurs and advances the ignition timing when knocking no longer occurs from the determined basic ignition advance angle. A knocking control method for an internal combustion engine. (2) A knocking control method for an internal combustion engine according to claim 1, wherein the learning retardation amount is calculated at every predetermined crank angle. (3) A knocking control method for an internal combustion engine according to claim 1, wherein the learning retardation amount is calculated at predetermined time intervals. (4) A knocking control method for an internal combustion engine according to any one of claims 1 to 3, characterized in that the learning retardation amount is set so that the corrected retardation amount falls within a predetermined range of values. .