JPS6250662B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention has a knock sensor that detects engine knocking, and determines whether knocking has occurred in the engine based on a signal from the knock sensor, and performs ignition timing control that feedback controls the ignition timing based on the determination result. Regarding the method. The behavior of knocking with respect to the engine's ignition timing is that if the ignition timing is advanced from top dead center, knocking will not occur up to a certain angle, and if the ignition timing is advanced further, the number of engine combustions (the number of ignitions) will increase.
The ratio of the number of knock occurrences to the number of knocks increases from 0 as the advance angle value increases. Generally, in knock feedback control, knocking is detected by a knock sensor, and the ignition timing is advanced or retarded so that the rate at which knocking is detected (knock occurrence rate) with respect to engine speed (number of ignitions) or time reaches a target value. and control. In the conventional knock feedback control method for ignition timing, knock detection is started after a predetermined starting condition is met, and if knocking is detected, the ignition timing is retarded by a certain angle, and the knock detection is delayed by a predetermined number of ignitions or by a predetermined number of times. If no knocking is detected after a period of time, the ignition timing is advanced by a certain angle. In this method, there are two types of feedback amounts: a constant advance angle and a constant retard angle. Since these feedback amounts do not slow down the control response speed, there is a limit to how small they can be. This conventional method has the following drawbacks. In other words, even if knocking is detected at a ratio to the target number of ignitions or time, in the conventional method, when knocking is detected, the ignition timing is always controlled in the retarded direction. It also moves. Furthermore, since the ignition timing fluctuates, the time interval between knocking occurrences or the number of ignitions from knocking occurrence to knocking occurrence varies greatly. The present invention aims to eliminate the drawbacks of the conventional method described above, and after a predetermined starting condition is established,
Measure the number of ignitions or time until knotting is detected and until knotting occurs,
The ignition timing feedback amount is determined stepwise or continuously according to the magnitude of this measured value.In other words, the relationship between the measured value and the feedback amount is that in the portion where the measured value is small, the amount of retardation is initially As the measured value increases, the retard amount decreases stepwise or continuously to 0, and as the measured value further increases, the feedback amount becomes an advance amount, and the feedback decreases stepwise from 0 to a predetermined measured value. A target time value corresponding to a knock detection ratio, or By setting the target ignition number value, the amount of feedback when knocking is detected at this time value and the ignition number value is 0, and the ignition timing is not moved unnecessarily. In other words, in the present invention, when knocking is detected at a time value corresponding to the target knocking detection ratio or at a time value close to the ignition number value, the amount of feedback is small, making it possible to reduce fluctuations in the ignition timing. It is something. Furthermore, since fluctuations in the ignition timing are smaller than in the conventional system, variations in the time interval from one knocking occurrence to the next knocking occurrence or the ignition frequency interval can also be reduced. Furthermore, since the stability of ignition timing can be improved, the amount of feedback can be reduced to 0 over a wide range of measured values of time or number of ignitions. In addition, in the present invention, if it is agreed that feedback will not be performed unless knocking occurs, if the ignition timing is in the ignition timing range where knocking does not occur,
Considering the problem that feedback does not occur and the ignition timing is not controlled to the advanced side, if knocking is not detected even after a predetermined number of ignitions or a predetermined time, knocking will be detected at the predetermined number of ignitions or a predetermined time. It is also characterized in that it is processed as if it were detected. As a matter of course, the calculation result of the feedback amount based on the measured value corresponding to the predetermined number of ignitions or the predetermined time is used to feedback control the ignition timing to the advanced side. In addition, in the present invention, in order to increase the response speed of feedback control and improve stability when the feedback target value is reached, the amount of feedback of the ignition timing according to the measured number of ignitions or the measured time is adjusted based on the past feedback history. It is characterized by changing the calculation method. In other words, if there has been a retard in the past and the current amount of feedback is a retard amount, this retard amount is increased more than normal to feed back, and if there has been an advance in the past and the current amount of feedback is a retard amount. If the amount of advance angle is the amount of advance angle, feedback is performed by increasing the amount of advance angle than usual, and if the target value of feedback is reached where advance angle and retardation continue alternately, feedback of retardation and advance angle is performed. Taking into account that the stability of ignition timing is better when the amount is smaller, when advance and retardation occur alternately, the amount of feedback is reduced compared to normal to perform feedback. It is a feature. An embodiment in which the present invention is applied to a four-cylinder engine shown in FIG. 1 and below will be described below. In FIG. 1, 1 is a 4-cylinder 4-cycle engine, 2 is a knock detector fixed to the engine 1 and detects the characteristic engine vibration when knocking occurs, 3 is a starter, and 31 is a starter switch. 4 is a rotation sensor that measures the rotation angle position of the engine 1, which generates a top dead center signal when the engine 1 rotates and reaches the top dead center position, and also divides one engine rotation equally from the top dead center position. A rotation angle signal is generated each time the crank angle is rotated by a certain crank angle (for example, 7.5 degrees in this embodiment). A pressure sensor 21 is built into the control computer 6, and pressure is transmitted from the intake manifold 12 of the engine 1 to a pressure input port through a pipe 13 to measure the intake manifold pressure. 5 is a known fuel supply device. 7,
Reference numeral 8 designates an igniter and an ignition coil as an ignition actuator using two coils and eliminating a distributor. The control computer 6 calculates the engine rotation speed from the time interval at which the rotation angle pulses generated by the rotation sensor 4 occur.
In addition, the intake manifold pressure is calculated from the output voltage of the pressure sensor 21 to measure the operating state of the engine, and a knock sensor (knock detector)
The knocking occurrence state is detected from the output signal No. 2, and the ignition timing is controlled. Furthermore, in order to control the ignition timing to a specific value when starting the engine, the voltage supplied from the starter switch 31 to the starter 3 is input to the control computer 6 as a starter signal. Further, in order to change the energization time of the ignition coil in accordance with the battery voltage, the battery voltage is taken into the control computer 6 as a battery voltage signal.
Reference numeral 15 denotes a power source that generates power at a voltage required by the control computer 6 from the voltage of a battery 19 mounted on the vehicle. Next, according to FIG.
will be explained in detail. 60 is the central processing unit (CPU) that calculates the ignition timing, and uses a 16-bit TMS9900 manufactured by TI. Reference numeral 70 indicates a read-only memory unit (ROM) that stores control programs and control constants, and 69 indicates a temporary memory unit (RAM) that is used to store control data while the CPU 60 is operating according to the control program. 61 is an interrupt control unit which notifies the CPU of the occurrence of an interrupt with three levels of priority. The interrupt factors in this embodiment include an interrupt caused by the generation of a rotation angle signal pulse from the engine rotation sensor 4;
There are three types of interrupts: a program interrupt which is caused by the CPU 60 itself setting the level of the program interrupt output port of the digital input/output port 64 to "0" level, and a timer interrupt which is caused by the timer every 8 milliseconds (ms). The timer section 62 is composed of a 16-bit counter that counts an 8 μs clock signal and a latch that stores and holds the counter value every time a rotation angle signal pulse of the rotation sensor 4 is generated. Therefore, in the interrupt processing for generation of rotation angle pulses, the CPU 60 reads the value of the crank angle counter section 63 to know the rotation angle position of the engine, reads the latch value of the timer section 62, and repeats this operation for two rotations. By determining the difference between the latch values at each angular position, it is possible to measure the time it takes for the engine to rotate between two rotational angular positions, and also to measure the engine speed. The crank angle counter section 63 is a counter with a lower digit in decimal notation and an upper digit in 4 decimal notation, and counts up based on the rotation angle signal of the rotation sensor 4, and when the next rotation angle signal is generated after the generation of the top dead center signal. The counter is reset to the value "0" and synchronized with the engine rotation. Therefore, the value of the crank angle counter is
By reading it out by the CPU, the engine rotational angular position can be determined in units of 7.5 degrees of crank angle. The digital input/output port 64 is a port used for inputting and outputting logic signals, and in order to recognize that the starter switch 31 is turned on when starting the engine, the digital input/output port 64 is a port used for inputting and outputting logic signals. Enter your level. It is also used to generate a program interrupt signal to be supplied to the interrupt control section 61. As is well known, the knock detection circuit 65 internally generates a knock detection pulse signal when the instantaneous value of the knock signal greatly exceeds the average level, and counts the number of pulses for a period specified by the CPU 60. The analog input port 67 is for measuring the voltage value of an analog signal, and is connected to the output voltage signal of the pressure sensor 21 that measures the intake manifold pressure of the engine 1 and the battery voltage for correcting the energization time of the ignition coil 7. Analog-to-digital (A/D) conversion of voltage. The energization ignition control section 68 energizes the two coil current actuator circuits of the igniter 8 and generates an ignition signal. This energization ignition control section 68
has several down counters, and is instructed by the CPU to determine which ignition coil 7 should be energized and ignited, and to determine the crank angle counter value and down count value at which the down counter should start counting. When the value becomes "0", the #1 coil energization signal and the #2 coil energization signal are set to level "0" for energization and to level "1" for ignition. 71 is a common bus,
The CPU carries control and data signals on this bus signal line to control peripheral circuits and send and receive data. The configuration, operation, and control method of the control program of this embodiment will be explained below. The control program has a priority relationship as shown in Table 1.

[Table] Consists of four layers of processing: three types of interrupt processing routines and a base routine that is constantly executed when no interrupt processing is performed. In the interrupt processing routine, processing that needs to be performed in synchronization with engine rotation and elapsed time is performed. First, the gear interrupt processing routine shown in FIG. 14 will be explained. The gear interrupt signal is generated every 7.5 degrees of crank rotation angle, with the top dead center position of cylinder #1 or #4 being 0 degrees. In the gear interrupt processing routine, first, in step 1100, the value of the crank angle counter is read from the crank angle counter section 63, and the rotational angular position of the engine 1 is determined. Next, in the energization ignition processing at step 1200, data is supplied to the energization ignition control unit 68 for control in accordance with energization ignition control data calculated by a program interrupt to be described later. Next, in step 1300, fixed position processing is performed to perform processing at a fixed angular position. FIG. 15 shows a flowchart of the energization ignition process in step 1200. First step 12
At 05, it is determined whether the engine has currently reached the angular position at which to start counting down to start energizing the #1 coil of the two ignition coils 7,
If the judgment is “YES”, step 1210 #1
At the same time as setting the initial value of the down counter for coil energization, down counting is started. In this embodiment, this down counter counts down the clock signal for 8 μs and when the value reaches "0", it sets the #1 coil energization signal to "0" and starts energizing. In step 1215, it is determined whether the engine has reached the angular position to start counting down for ignition of the #1 coil, and if the determination is "Yes",
At step 1220, the initial value of the down count for energizing the #1 coil is set, and at the same time, the down count is started. In this embodiment, this down counter counts down with an 8 μs clock signal, and when the value reaches "0", stopping the energization of the ignition coil corresponds to ignition, so the energization signal of the #1 coil is When set to logic "1", energization of ignition coil #1 is stopped. Steps 1225 and 12 for #2 coil as well.
Similarly, energization ignition control is performed at 30, 1235, and 1240. Next, the fixed position processing in step 1300 will be explained. In the energization ignition process, the angular position at which the process effectively operates varies depending on the value calculated by the program interrupt, whereas in the fixed position process, the process is performed at a predetermined angular position. FIG. 4 shows at which rotational angular position the fixed position process 1300 is performed. Crank angle 0~60
Measure the rotation time to rotate a 60 degree crank angle at 180 to 240 degrees. Measure the intake manifold pressure at 0 and 180 degrees. Angle 22.5~60 degrees,
Count the knock detection pulses from 202.5 to 240 degrees. Also, a program interrupt is generated at 60 and 240 degrees, and the program interrupt processing routine is activated. 16th
A flowchart of this fixed position processing 1300 is shown in the figure. In step 1310, the crank angle is determined,
Perform TDC treatment 1320 at 0, 180 degrees, 22.5 ~
Perform 22.5 degree processing 1330 at 200.5 degrees, 60 degrees,
Perform 60 degree processing 1340 at 240 degrees, and do nothing else. FIG. 17 shows a flowchart of the TDC processing 1320. In step 1321, the timer section 6
The timer latch value of step 2 is taken in as TP1, and then in step 1321, the A/D converted value of the intake manifold pressure Pm is taken in. FIG. 18 shows a flowchart of the 22.5 degree processing 1330. In step 1331, the signal port of the digital input/output port 64 is set to "0". As a result, the knock detection circuit 65 starts counting knock detection pulses. FIG. 19 shows a flowchart of the 60 degree process 1340. At step 1341, the timer latch value is fetched and set as TP2. In step 1342, the 60 degree crank angle rotation time Tm is determined using equation (1). Tm=TP2-TP1...(1) Digital input/output port 6 in step 1343
The signal output port No. 4 is set to level "1" and counting of the knock detection pulses is completed. Further, in step 1344, the program interrupt output port of the digital input/output port 64 is set to level "0" and then set to level "1" again to generate a program interrupt signal and start the program interrupt processing routine. FIG. 20 shows a flowchart of program interrupt processing. At step 2100, the ignition timing to be applied to the current operating condition is calculated based on the feedback results calculated in the base routine described below. Actual control of the energization ignition control section is performed by the gear interrupt processing routine as described above. In step 2200, it is determined whether the feedback flag _fFB is "1". When the flag _fFB is "1", the start of feedback is instructed in the check evaluation process at step 2300, and
Since this means that feedback processing such as map rewriting, which will be described later, has not been completed, the program interrupt processing is ended without performing the knock evaluation in step 2300. When the flag _fFB is not "1", the value is "0", and at this time, the start of feedback has not been instructed and the feedback processing is not in progress, so step 2300 is executed.
Performs nok evaluation processing. Flag f _FB is “1”
The reason why nok evaluation is not performed at this time is because even if nok evaluation is performed while map rewriting is in progress, the control conditions are not clear and the feedback method becomes complicated, so feedback control is not performed. This is because there is a risk of impairing the stability of In the knock evaluation processing at step 2300, it is determined whether or not to start feedback processing based on the count value of the knock detection pulses of the knock detection circuit 65 and the number of firings in which no knocks are detected, and the base routine processing actually performs the feedback processing. Activate the routine. To activate, set flag _fFB to “1”. Here, a method of calculating ignition timing in this embodiment will be explained. FIG. 3 shows a graph of a torque curve with respect to ignition timing at a certain engine speed. Ignition timing is expressed in angle before top dead center (degrees BTDC), and the larger the value, the more advanced the ignition timing is. Torque curve 1 indicates the throttle valve is close to fully open, and MBT1 is the MBT of torque curve 1.
(MINIMUM ADVANCE FOR BEST
TORQUE), and the ignition timing at which knock starts to occur in this case is indicated as knock limit 1. Ignition using MBT is best from the standpoint of engine efficiency and fuel efficiency, but when knocking occurs, it has a negative effect on the engine itself, and the unique sound generated is not good for the driver's hearing, so it is necessary to reach the knocking limit or the knocking. It is customary to set the ignition timing slightly behind the limit. Torque curve 2 shows the relationship between torque and ignition timing when the throttle valve is in a half-open state. Display MBT and Knock Limit as MBT2 and Knock Limit 2. In this case, the knock limit is on the advance side of the MBT, so it is best to ignite at the MBT. Therefore, in this embodiment, MBT is set as the maximum advance ignition timing, and feedback control is performed from this value to the retard side using the output signal of the knock sensor. The ignition timing calculation formula is shown in equation (2). θx=θ _PP -θ _FM -PXθ _T ...(2) θx: Final ignition timing (degrees BTDC) θ _PP : Fixed map advance angle θ _FM : Feedback map value P: Learning coefficient θ _T : Feedback retard amount Hereinafter, how to obtain each term in equation (2) will be explained according to the flowchart of the ignition timing calculation process 2100 in FIG. 21. In step 2105, the engine rotational speed N is determined from the 60 degree crank angle rotation time Tm calculated in the gear interrupt process using equation (3). N (rpm) = 10 ⁷ × 1/Tm (microseconds) ... (3) In step 2110, the A/D conversion value Vpm of the intake manifold pressure taken in by the gear interrupt processing is
Find the intake manifold pressure from equation (4). Pm (mmHg) = A ₁ × Vpm + A ₂ ... (4) A ₁ , A ₂ : Conversion coefficient In step 2115, it is determined whether the starter is activated when starting the engine, and if "YES", the ignition timing is changed to 7.5 in step 2120. degree, energizing time 11.5
Set it to a fixed value of milliseconds and proceed to step 217.
Proceed to step 5. If the starter is not operating, the process advances to step 2125. The fixed map advance angle (θ _PP ), the feedback map value (θ _FM ), and the number of feedbacks (N _FB ) for determining the learning coefficient P are calculated using the engine speed N and intake manifold pressure (Pm) as two axes. Fixed advance angle map (θ _PP map, see Figure 5), Feedback map (θ _FM map, see Figure 6)
(See figure) It is determined from the feedback frequency map ( _NFB map, see Figure 7). The θ _PP map is a fixed data map that stores MBT, and the θ _FM map and N _FB map are variable data maps that are rewritten by feedback or the like as control progresses. The θ _FM map stores the amount of retardation from the MBT for each operating state classification, and the N _FB map shows how much each value in the θ _FM map has been rewritten as a result of past feedback. The horizontal axis of the θ _PP map in Figure 5 is the engine speed N
In the low rotation range below 1000 rotations, the increment of the engine rotation speed N is small with respect to the column number X corresponding to the horizontal axis, and there is no overall proportional relationship (single linear function type). The vertical axis of the θ _PP map is the intake manifold pressure Pm, and the line number Y corresponding to the vertical axis
The relationship is proportional to (single linear function type). The θ _FM map in Fig. 6 and the N _FB map in Fig. 7 are defined only for the high load region of the θ _PP map in Fig. 5, but for the defined region, engine rotation, intake manifold The relationship between pressure, column number X, and row number Y is the same as the θ _PP map. θ _PP map, θ _FM map, N _FB map to θ _P
When determining the values of _P , θ _FM and N _FB , use X and Y. Therefore, in step 2125 of FIG. 21, the rotation speed N is converted to the column number X according to equation (5). Xi is a column number and is an integer, and the decimal point of X indicates the middle rotation speed between the rotation speeds specified on the horizontal axis of the map. In step 2130, the intake manifold pressure (Pm) is converted to the row number Y according to equation (6). Yi is a row number and an integer. The decimal point part of Y indicates the intake manifold pressure that is between the intake manifold pressures specified on the vertical axis of the map. In step 2135, the fixed map advance angle θ _PP is set as
Read from the fixed advance angle map (θ _PP map) at X and Y obtained in steps 2125 and 2130, (7)
It is obtained by performing four-point interpolation calculation using the formula. In step 2140, a feedback map value θ _FM is calculated by four-point interpolation from the feedback map (θ _FM map) in X and Y using equation (8). In step 2145, the number of feedbacks N _FB is calculated by four-point interpolation from the feedback number map (N _FB map) in X and Y using equation (9). Each value of the feedback count map is the same
This shows how much the value of the θ _FM map of the Y number has been feedbacked in the past, and the procedure for rewriting each value of the feedback map (θ _FM map) and the number of feedback map (N _FB map) will be described later. This is explained in the section on base routines. In step 2150, the number of feedback N
Learning coefficient map (P map) as shown in Figure 9 on _FB
Search for learning coefficient P. As is clear from the learning coefficient map (P map) in FIG. 9, the learning coefficient P becomes smaller as the number of feedback increases. This is because each map value of the θ _FM map is considered to approach the feedback target value as feedback is performed, and the amount of feedback is reduced to improve control stability. In step 2155, the feedback angle _θFB is determined using equation (10). θ _FM = θ _FM + P x θ _x ...(10) The term P x θ _T exists because the distribution of engine speed and intake manifold pressure during the knock evaluation period, such as during acceleration and deceleration, as will be described later. This is because if θ FM is not within a certain range, there is no feedback to the feedback map value _θFM , so this term provides feedback. In order to prevent the ignition timing from advancing beyond MBT, it is checked in step 2160 that _θFB is not negative, and if it is negative, it is set to 0 in step 2165. In step 2170, the final ignition timing θ _x is set as (11)
Calculate by formula. θx=θ _PP −θ _FB (11) In step 2175, in the case from step 2170, calculation is performed using an 8ms interrupt processing routine to be described later. Calculate the coil energization angle from the coil energization time T _ON based on the engine speed, and set the final ignition timing θx.
Find the energization start angle θ _ON from coils #1 and #2.
The starting angular position of the down-count for energization and the initial value to be set to the down counter, and the starting angular position of the down-count for ignition and the initial value to be set to the down-count are calculated. FIG. 22 shows the knock evaluation processing step 230.
0 is shown. In step 2305, the knock interval, which is a software counter of the number of ignitions since the start of knock evaluation in the base routine described later, is set.
Increase WNO by 1. In steps 2310 and 2315, it is determined whether the engine speed N and intake manifold pressure Pm are in a steady state within a certain difference △N, △Pm from the comparison reference values Ns, Pms, respectively, and the answer is "NO".
If so, in step 2325, the steady state flag _fST is set to the value "0". If it is in a steady state, in step 2320, the average value of the column number X corresponding to the engine rotation speed N and the row number Y corresponding to the intake manifold pressure Pm during the knock evaluation period is calculated, so the integrated value Σ _X of X and Y is calculated.
Σ Add the current X and Y to _Y. In step 2335, 22.5
60 degrees and 200.5 to 240 degrees, the knock pulse count value Nkpls is taken from the knock detection circuit 65 that counts the knock detection pulses, and in step 2340
It is determined whether the value of Nkpls is larger than the comparison reference value kp, and if "YES", it is assumed that a knock has been detected and the feedback flag _fFB is set to "1" in step 2350. If “NO” in the judgment step 2340, the number of ignitions after the start of knock evaluation in step 2345 is equal to the predetermined number of ignitions Nmax (for example,
100 ignitions) has elapsed, and if it has elapsed, the feedback start flag _fFB is set to "1" in step 2350, and the base processing routine starts feedback. Step 2
If the determination at step 345 is "NO", no action is taken and the knot evaluation process is ended. FIG. 23 shows a flowchart of the 8 millisecond (ms) interrupt processing routine. In step 3100, the battery voltage V _B is taken from the analog input port 67, and in step 3200, the battery voltage V _B is taken in.
According to the energization time characteristics shown in FIG. 13, the coil energization time T _ON is calculated. Next step 33
00, 1 is added to the counter _CIM on the software for creating a time signal every minute. FIG. 24 shows a flowchart of base processing. In step 4700, initialization processing is performed to clear the θ _FM map and N _FB map, clear counters on the software, and set initial values of parameters and the like. In step 4200, interrupts are enabled. In step 4300, it is determined whether the feedback flag _fFB , which is set by the program interrupt processing and instructs the start of feedback processing, is "1". If "Yes", the feedback processing is performed in step 4400. . “NO”
If so, proceed to step 4500 without doing anything. In step 4500, it is determined whether the 1-minute software counter C _IM is greater than K _IM (=60 seconds/8ms=750), and if ``YES'', 1 minute has passed, so the process is forgotten in step 4600. Process. FIG. 25 shows a flowchart of the feedback processing 4400. In this process, the θ _FM map feedback does not immediately feed back the map feedback data generated from the data of a certain knot evaluation period, but feeds back to the map after the next knot evaluation period ends, and is delayed by one knot evaluation period. Let's do it. This is because the feedback angle θ _{FB is}
It is expressed as = θ _FM +P×θ _T , and as described later, θ _FM
This is because approximately an amount of P×θ _T is fed back to the θ FM map, so the feedback angle θ _FB is approximately twice as much as when it is not fed back to the θ _FM map. In step 4405, the operating state during the knock evaluation period is determined by the average value of the column number X and line number Y used in the ignition timing calculation 2100 of the program interrupt processing using equations (12) and (13). = ΣX / WNO ... (12) = ΣY / WNO ... (13) ΣX: Accumulated value of X during the knock evaluation period ΣY: Accumulated value of Y during the knock evaluation period (number of ignitions) θ The point at coordinates (,) on _{the FM} map is used as the feedback center when performing map feedback. In step 4410, the center of the operating state (,
) is within the defined feedback region of the θ _FM map as shown in Figure 6,
If "YES", the feedback amount θ _T is calculated in step 4415. In step 4425, it is determined whether the steady state flag _fST is "1", and if it is "YES", the map feedback flag ( _fMFB ) is set to "1" in step 4430. If "NO", the process advances to step 4435, and the map feedback flag ( _fMFB ) is set to "0". Therefore, in this case, the feedback amount _.theta.T is calculated, but it is not fed back to the map because the map feedback flag _fMFB , which indicates whether map feedback is to be performed, is not "1". If it is determined in step 4410 that it is not within the feedback region, θ _T is set to "0" in step 4420, and the process proceeds to step 4435, in which the flag _fMFB is set to 0. In step 4440, it is determined whether the value of the map feedback flag 2f _MFB2 to which the value of the flag f _MFB of the previous feedback processing has been transferred is "1". If "YES", the map feedback flag is transferred in step 4445. Process. At this time, the data used for map feedback is the data calculated in the previous feedback process. Step 44
If the determination at step 44 is "NO", the map feedback is not performed and the process proceeds to step 4450. In step 4450, the map feedback data calculated this time is set as a parameter for map feedback in the next feedback process. ₂ = X coordinate of the center of the feedback used for map feedback ₂ = Y coordinate of the center of feedback used for map feedback θ _T2 : Feedback amount used for map feedback f _MFM2 : Map feedback flag 2 Step At step 4455, processing for starting the knot evaluation is performed. That is, the knock interval
WNO, ΣX, ΣY are cleared to "0", feedback flag f _FB is set to "0", steady state flag f _ST is set to "1", steady state judgment comparison reference value Ns, Pms is the current engine rotation speed N, intake manifold pressure
Configure Pm settings. FIG. 26 shows a flowchart of the feedback amount θ _T calculation process in step 4415. In this embodiment, the feedback amount θ _T is determined by the knock interval WNO, which is the number of ignitions from the start of the knock evaluation to the end of the knock evaluation. The knock evaluation period is determined by whether a knock is detected or the number of aborts.
It ends when Nmax has passed. FIG. 8 shows a graph of the relationship in determining the feedback amount θ _T from the knock interval WNO. _θT
When the value of θ T is positive, it is feedback toward the retard side, and when the value of θ _T is negative, it is feedback toward the advance side. Basically, the retard side is a straight line a ₂ and the advance side is a straight line b ₂
determined by the graph of Actually, the straight lines a ₂ , c,
The values of WNO and θ _T at the end points of b ₂ are stored as a basic feedback amount θ _f map, and θ _f is obtained by linear interpolation calculation from this map using the knock interval WNO. This θ _f is multiplied by a correction coefficient, which will be described later, to obtain θ _T. Also, when a large angle must be retarded, the larger the value of the _a2 part of the graph in Figure 8, the faster the response to the retard side, and when a large angle must be advanced, The larger the value in the _b2 part of the graph, the faster the response to the advance angle side. Therefore, in this example, if the retardation continued in the past,
If the retard angle characteristic was changed from a ₂ to a ₃ and the advance angle continued in the past, the advance angle characteristic was changed from b ₂ to b ₃ to speed up control responsiveness. Further, when the feedback target value is reached where the advance angle and the retard angle continue alternately, the stability of the ignition timing is improved by reducing the amount of feedback of the retard angle and the advance angle. Therefore, in this embodiment, when the advance angle and the retard angle continue alternately, the retard angle characteristic is changed from a ₂ to a ₁ , and the advance angle characteristic is changed from b ₂ to b ₁ to improve control stability. . This will be explained using the flowchart of the feedback amount θ _T calculation process 4415 shown in FIG. 26. In step 4415-1, the knock interval
Find θ _f from the θ _f map using WNO by performing interpolation calculations. In step 4415-2, it is determined whether the steady state flag f _ST managed by the program interrupt processing is "0" or "1", and if it is unsteady at "0", the retardation characteristic is determined in steps 4415-3 and 4415-4. Both the parameter ia and the lead angle characteristic parameter ib are set to 0, and in step 4415-5, the basic feedback amount θ _f is directly set as the feedback amount θ _T . If it is determined in step 4415-2 that the value of the steady state flag f _ST is "1" indicating steady state, the value of the current feedback amount θ _f is checked in step 4415-6. If θ _f >0 and the angle is retarded, the process proceeds to step 4415-7, where the value of _fl indicating the previous feedback history is determined. When f _l > 0, the previous time was a retard. When f _l < 0, when the lead angle was f _l = 0. There was no lead or retard last time. If f _l > 0 and the previous time was also a retard, step 441.
Determine the value of the advance and retard characteristic parameter ia in step 5-8. ia = 1: Retard angle characteristic is a in Fig. 8. ₃ ia = 0: Retard angle characteristic is a in Fig. 8. ₂ ia = -1: Retard angle characteristic is a in Fig _{. 8.} If ia≦0, step 2223-9 to ia 1
is added to enhance the retardation characteristic, and step 4415
Proceed to -12. If ia > 0, the retard characteristic cannot be further enhanced and the process continues in step 4415-
Proceed to step 12. If _fl = 0 as determined in step 4415-7, the process directly advances to step 4415-12. If f _l <0 in the judgment at step 4415-7, that is, the previous angle was advanced and the current angle is retarded. In this case, the value of the advance angle characteristic coefficient ib is determined at step 4415-10, and ib If ≧0, subtract ib by 1 in step 4415-11 and proceed to step 44.
Proceed to 15-12. In step 4415-12, the feedback amount θ _T is calculated using equation (15): θ _T =θ _f ×2 ^ia (15). The process then proceeds to step 4415-20. Step 4415-
If θ _f =0 in step 4, set θ _T =0.
Proceed to 15-20. In step 4415-6, if θ _f <0, that is, the current feedback amount is on the advance side, the process advances to step 4415-13. Step 4415-1
In step 3, if _fl > 0 and the previous feedback is retarded, the process advances to step 4415-14. If the retard characteristic coefficient ia is ia≧0 in step 4415-14, 1 is subtracted from ia in step 4415-15. The process then proceeds to step 4415-18. If ia<0 in step 4415-14, the process proceeds to step 4415-18 without doing anything. In step 4415-13, if f _l =0 and the advance/lag amount in the previous feedback is 0, the process immediately proceeds to step 4415-18. In step 4415-13, if f _l <0 and the previous feedback was advanced, step 4415
Proceed to -16 and check the value of the advance angle characteristic coefficient ib. ib
If ≦0, add 1 to increase the advance angle feedback amount. The process then proceeds to step 4415-18. If ib>0 in step 4415-16, the advance angle feedback amount cannot be increased any further, so the process directly advances to step 4415-18. In step 4415-18, the amount of feedback is calculated using the following equation. θ _T ← θf×2 ^ib (16) Then, the process proceeds to step 4415-20. In step 4415-20, the feedback amount θ _T calculated this time is transferred to _fl . FIG. 27 shows a flowchart of the map feedback processing step 4445. In this embodiment, the map feedback includes the previous average value ₂ of column number X corresponding to the rotation speed N during the knock evaluation period, and the previous average value ₂ of row number Y corresponding to the intake manifold pressure Pm. Feedback is performed for each element of the θ _FM map within two columns or two lines from the feedback center ( ₂ , ₂ ), which is defined as two coordinates. In steps 4445-1 to 4445-4 _, the lower limit value X of the column number
_L , upper limit value X _H , lower limit value Y _L of line number Y, upper limit value Y _H
is obtained using equations (17) to (20). The integer part of _L = ( ₂ - 1) or ₂ - 2 ( ₂
is _an integer value ₎ or X _FML (when the _above value is smaller than _the lower _{limit of}
(when is an integer value) or Y _FML (when the above _value is smaller than the Y lower limit Y _FML of the θ _FM map) ... _{( 18} ) _is larger than _the upper limit _{of X of the} θ _FM map ) ...(20) Column number of the element to be rewritten in step 4445-5
Set X _L to Xj and Y _L to line number Yj. In step 4445-6, the feedback center ( ₂ ,
₂ ) Find the X-direction deviation △X and Y-direction deviation △Y between points (Xj, Yj). However, △X and △Y are obtained by multiplying the actual deviation by 4 to include only the integer part. In step 4445-7, the distance L map is searched using ΔX and ΔY to find the distance l. 10th
An example of a distance map is shown in FIG. In the example of FIG. 10, the distance l increases with equal weight for ΔX and ΔY. In example 2 of FIG. 11, the distance increases with ΔX weighing twice as much as ΔY. The amount of feedback decreases as the distance increases, so the first
In the case of Example 2 in FIG. 1, the amount of feedback decreases rapidly as it moves away from the feedback center in the ΔY direction, that is, in the intake manifold pressure direction. Step 4445-8 searches the feedback rate k from the feedback rate map (K map) using the distance l. The relationship between the distance l and the feedback rate k is shown in the characteristics of the K map in FIG. In step 4445-9, the number of feedbacks N _FB (Xj, Yj) of the N _FB map is searched. In step 4445-10, the learning coefficient P is searched from the P map shown in FIG. 9 using N _FB (Xj, Yj). In step 4445-11, it is determined whether the previous feedback amount θ _T2 is positive or negative, and if θ _T2 ≧0,
In step 4445-13, θ _FM is calculated using equation (21).
Rewrite (Xj, Yj). θ _FM (Xj, Yj) = minimum value {θ _PP (Xj, Yj), θ _FM (Xj, Yj) + K × P × θ _T2 } ... (21) If θ _T2 < 0, in step 4445-12 Rewrite θ _FM (Xj, Yj) using equation (22). θ _FM (Xj, Yj) = maximum value {0, θ _FM (Xj, Yj) + K×P×θ _T2 } ...(22) After that, in step 4445-14, the rewriting element θ _FM ( _Xj , Feedback frequency N corresponding to Yj) _FB map element N _FB (Xj, Yj)
Rewrite according to equation (23). N _FB (Xj, Yj) = N _FB (Xj, Yj) + K
...(23) This allows the element θ _FM (Xj, Yj) of the θ _FM map
Memorizes how much feedback has been received. In step 4445-15, 1 is added to Xj to change the position of the rewritten element. In step 4445-16, the magnitude relationship between Xj and _XH is determined, and if _XjXH , step 4445-16 is performed.
Proceed to step 6 to continue the rewriting operation. If Xj> _XH , set Xj to _XL in step 4445-17, and add 1 to Y in step 4445-18. Then, in step 4445-19, the magnitude relationship between Yj and _YH is determined, and if _YjYH , step 4445-6
Go to and continue the rewrite operation. If Yj> _YH , map feedback processing ends. In the above-described map feedback processing step 4445 of FIG. 27, the range and distribution of feedback to the θ _FM map are determined by the deviation ΔX between the feedback center ( ₂ , ₂ ) and the column number X of the rewritten element (Xj, Yj). Determining the distance l from the deviation ΔY of the row number Y has the following advantages. The intake manifold pressure and engine speed corresponding to each row and each column of the θ _FM map and the θ _PP map are the same for the sake of computational advantages. In this embodiment, the θ _PP map stores the MBT, but in the low-speed portion where the value changes suddenly, the difference between the engine speeds in adjacent columns is reduced to accurately represent the MBT. In such an operating range, the feedback control value also fluctuates rapidly, so rather than determining the amount of feedback for that element based on the difference in engine speed from the feedback center, it is better to determine the amount of feedback based on the deviation of the column number. Harmonizes with the map structure. In addition, in this map feedback processing step 4445, the feedback center (
₂ , ₂ ) and the rewriting element θ _FM of the θ _FM map
Find the distance l from the distance map (L map) using the X direction deviation △X and Y direction deviation △Y, which are obtained by multiplying the deviation of the actual coordinates of (Xj, Yj) by 4 to make only the integer part, and
The feedback coefficient k is searched from the feedback coefficient map (K map). Therefore, 1/4
Feedback coefficients can be specified for deviations for each column number and row number. Feedback performance is improved by being able to specify the feedback amount distribution in such a fine manner. FIG. 28 shows a flowchart of forgetting processing step 4600 in the base processing of FIG. 24.
In this embodiment, feedback control is performed to retard the angle when the knock interval is small and to advance the angle when it is large, and the control speed to the advance side is slower than the control speed to the retard side. Furthermore, the values of each element in the operating state category that is not used for a long time in the θ _FM map may be far away from the current feedback target value. Therefore, the feedback target value may be reached more quickly if the value of each element is forgotten than if it is left as is. Therefore, in this feedback, since the control speed to the retard side is fast, all elements of the θ _FM map are subtracted by a predetermined amount every minute to be used as a forgetting effect. In step 4605, it is determined whether the subtraction amount calculation completed flag _FIM is set to "1". If it is "0", the subtraction amount has not been calculated yet, so step 4610 is performed.
Then, the subtraction amount θ _fgt is calculated according to equation (24), and in step 4615, 1 is added to the software elapsed time T in minutes. θ _fgt = θ _FGT / (T-1) ... (24) θ _FGT : Initial amount to be subtracted T: Elapsed time In step 4620, the column number Xk of the element of _{the θ FM map} is rewritten for forgetting. Set the X lower limit value X _FML , and set the Y lower limit value Y _FML of the θ _FM map in the row number Yk. Then, in step 4625, the flag _FIM indicating that the amount of subtraction has been calculated is set to "1". Then, in step 4630, the element θ _FM (Xk, Yk) of the θ _FM map is subtracted by θ _fgt , and if θ _FM (Xk, Yk) < 0 in step 4640, θ _FM (Xk, Yk) is subtracted in step 4645. Set to 0. At step 4650, Xk is added by 1, and at step 4655, Xk and θ _FM map's X upper limit X _FMH
If Xk≦ _XFMH , this forgetting process is terminated, and the next time the forgetting process is executed, the forgetting operation is performed on the next element of the θ _FM map. If Xk > X _FMH in step 4655, set X _FML to Xk in step 4660, and then
At 665, add 1 to Yk. Step 467
If Yk is 0 and Y _FMH is compared in size, if Yk≦Y _FMH , this forgetting process is terminated and the next time the forgetting process is executed, the forgetting operation is performed on the next element of the θ _FM map. It will be done. If Yk> _YFMH in step 4670, step 4675 subtracts a one-minute software counter _CIM by a time _KIM corresponding to one minute of elapsed time. Furthermore, in step 4680, the subtraction completed flag _FIM is set to "0". Therefore, when the value of K _IM is added to C _IM again, the next forgetting process is started. In the embodiments described above, the amount of feedback is determined depending on the number of ignitions from the start of knock evaluation until a knock is detected. An example will be described in which the amount of feedback is determined depending on the magnitude of . Only changes from the previously described embodiments will be described. In step 2345 of the knock evaluation routine 2300 in FIG. 22, the elapsed time WT from the start of the knock evaluation and its comparison value WTmax are compared with WTWTmax. The 8ms interrupt processing routine in FIG. 23 is changed to step 3400 as shown in FIG.
Add ~3500. That is, during the knock evaluation period in which the feedback flag _fFB =1 does not hold, the process advances to step 3500 and 1 is added to the elapsed time WT. Furthermore, in step 4455 of the feedback processing (FIG. 25) in the base processing routine,
The check evaluation start process is performed, but instead of clearing the check interval WNO to “0”, the elapsed time is
Clear WT to “0”. Therefore, WT indicates the elapsed time from the start of the knock evaluation in units of 8 ms. In addition, in the graph of the relationship between knock interval WNO and feedback amount _θT in Figure 8, the horizontal axis is WNO
Configure by changing from to elapsed time WT. The basic feedback amount map (θ _f map) also shows the elapsed time.
WT and feedback amount θ _f are memorized. 26th
Feedback amount θ _T calculation routine 4415 in the figure
In step 4415-1, the basic feedback amount θ _f is calculated by interpolation from the θ _f map using the WT. By doing this, the amount of feedback can be determined based on the length of time from the start of the knock evaluation until the knock is detected. As explained above, the present invention has the following features and advantages. (1) Depending on the number of ignitions or the length of time from when the predetermined measurement start conditions are met until knocking occurs (that is, the knocking interval),
By determining the amount of feedback, the stability of ignition timing can be increased. (2) By changing the method of determining the amount of feedback from the knock interval based on the past feedback history, that is, a signal that determines whether feedback control was being performed in the retard direction or in the advance direction, the advance angle can be adjusted. ,
Control response in both directions of retardation can be made faster.

[Brief explanation of the drawing]

FIG. 1 is a configuration diagram showing one embodiment of the present invention, and FIG.
The figure is a block circuit diagram of the control computer shown in Figure 1, and Figure 3 is an ignition timing diagram for explaining the present invention.
Torque characteristic diagrams and Figures 4 to 13 respectively serve to explain the operation of the embodiment shown in Figure 1. Figure 4 is an engine crank angle-fixed position processing relationship diagram, Figure 5 is a fixed advance angle map, and Figure 5 is a diagram showing the relationship between engine crank angle and fixed position processing. Figure 6 is a feedback map, Figure 7 is a feedback number map, Figure 8 is a knock interval-feedback amount characteristic diagram, Figure 9 is a learning coefficient map characteristic diagram, Figure 10 is a distance map example 1, and Figure 11 is a characteristic diagram of a learning coefficient map. is distance map example 2,
Figure 12 is a feedback rate map characteristic diagram,
3 is an energization time map characteristic diagram, FIGS. 14 to 28 are flowcharts of the control program of the embodiment of FIG. 1, and FIG. 29 is a flowchart of the main part of the control program of another embodiment of the present invention. It is. 1...Engine, 2...Knock sensor, 6...
control computer.

Claims (1)

[Claims] 1. A knock sensor for detecting engine knocking, and a signal from the knock sensor to determine whether knocking has occurred in the engine;
The ignition timing control method performs feedback control of the ignition timing based on the knocking determination result, the ignition timing control method comprising: measuring the number of ignitions or the time from when a predetermined measurement start condition is met until the knocking occurs; and the ignition timing determining whether the ignition timing is being feedback-controlled in the advance direction or in the retard direction based on the past value of the feedback amount that determines the ignition timing; 1. An ignition timing control method characterized by calculating a feedback amount for determining ignition timing based on the ignition timing, and controlling the ignition timing according to this feedback amount. 2 If the knocking does not occur until a predetermined number of ignitions or a predetermined time after the measurement start condition is met, calculate the amount of feedback by assuming that knocking has occurred after the predetermined number of ignitions or a predetermined time. An ignition timing control method according to claim 1, characterized in that: