JPH086630B2 - Engine controller - Google Patents

Description

The present invention relates to an engine control device, and more particularly, to detecting the combustion state of each cylinder of the engine and performing learning control for each cylinder so that the combustion state becomes a predetermined state. The present invention relates to the engine control device.

(Prior Art) Conventionally, as a control for controlling the combustion state of an engine to a predetermined state, for example, there is provided a sensor for detecting knocking of the engine and retarding the ignition timing when the knocking is detected. well known. Further, in such a mere feedback control that controls only after knocking occurs, since a delay is inevitably generated between them, it is desired to eliminate the delay during this period, for example, in JP-A-58-222976. As described, the learning correction value for each operating state of the engine is stored in each area of the memory partitioned by the engine speed and the load, and the learning correction value is stored the next time the operating state occurs. It is known that the ignition timing to be finally controlled is determined by calculating a correction value and a basic advance value. According to such knocking learning control, the optimum ignition timing can be obtained in response to a change in the operating state. However, in a multi-cylinder engine, there are inevitable variations in air-fuel ratio, filling amount, combustion chamber temperature, and compression ratio among the cylinders. Therefore, if feedback is uniformly applied to all cylinders as described above, When looking at individual cylinders, there is a problem that optimum control is not being performed.

Taking into account the variation between cylinders, the ignition timing corresponding to the knocking occurrence limit is determined for each cylinder, as described in JP-A-58-192968. Is simply to set the ignition timing for each cylinder, and is not to control the ignition timing to be optimum in accordance with changes in the operating state.

Further, based on these conventional techniques, if feedback control is performed for each cylinder and correction is performed with a learning value for each cylinder, more accurate control can be performed. In this case, for example, in the case of four cylinders, if the memory for storing the learning value is required for four, the overall capacity of the memory becomes large. In other words, if the same data size is used for each cylinder, the RAM capacity will increase and it will be necessary to add more RAM, leading to a problem of cost increase. Furthermore, in a general engine, a one-chip CPU is used. Since this one-chip CPU has built-in ROM and RAM, adding RAM means that the bus line is taken out of the CPU, and noise resistance Will be worse. In addition, when the 1-chip CPU is used as an external RAM, the bus terminals are used as I / O ports in the single port mode, which is the normal usage pattern. Will need to be added.

(Object of the Invention) In view of the problems of the engine control device according to the related art as described above, the present invention provides an engine control device in which the control amount is corrected by a learning correction value for each operating state of the engine. The purpose is to reduce the overall storage capacity without degrading the accuracy of each cylinder.

(Structure of the Invention) According to the present invention, even if the combustion state of the engine, for example, the knocking state of the engine varies greatly depending on the season and the octane number of gasoline, the difference in knock limit between the cylinders is reduced. Since it does not change so much, it is actually a problem if the learning correction value is stored as it is for a specific cylinder and the learning correction value is stored as a difference from that of the specific cylinder for other cylinders. It is based on the finding that there is no such decrease in accuracy and the capacity of the memory can be reduced, and its overall configuration is as shown in FIG. That is, the engine control device according to the present invention includes an operating state detecting means for detecting an operating state of the engine, a basic control amount determining means for determining a basic control amount based on an output of the operating state detecting means, and an engine Cylinder combustion detection means for detecting the combustion state for each cylinder, feedback correction amount determination means for determining a feedback correction amount for each cylinder according to a signal from the cylinder combustion detection means, and a learning value based on the feedback correction amount. Learning value calculation means for calculating, cylinder identification means for identifying the current cylinder, first learning value storage means for storing a learning value of a specific cylinder in response to a signal from the cylinder identification means, and learning for other cylinders. Difference calculation means for calculating the difference between the value and the learning value of the specific cylinder, second learning value storage means for storing the output of the difference calculation means as the learning value of the other cylinder, and basic control A final control amount determining means for determining the final control amount for each cylinder from the feedback correction amount and the learned value, and combustion control means for controlling the combustion state of the engine by the output of the final control amount determining means. Are stored as they are, and the learning values of the other cylinders are stored as differences from the learning values of the specific cylinder. The operating state detecting means detects the operating state of the engine, for example, from the output of a crank angle sensor attached to a distribution machine (distribulator). Further, the combustion control means controls the combustion state of the engine by controlling a fuel injection device and an ignition device, for example.

(Operation) The operating state detecting means determines the operating state of the engine based on the engine speed and the load. The basic control amount determining means receives the output of the operating state determining means and calculates a basic control amount, for example, a basic ignition timing, from a map for each operating region. Also,
The feedback correction amount determination means receives a signal from the combustion detection means provided for each cylinder and corrects the feedback correction amount for each cylinder. Furthermore, the learning correction value is calculated by the learning value calculation means based on the feedback correction amount, and the learning correction value for each operating region is stored and updated. The final control amount determining means calculates the final control amount from the basic control amount, the feedback correction value and the learning correction value, and outputs a control signal for the combustion control device of the engine, for example, the ignition device or the fuel injection device. In this way, the combustion state of the engine is controlled. By the way, the storage and updating of the learning correction value by the cylinder identifying means, the difference calculating means, and the first and second learning value storage means in the present invention are performed as follows. That is, if the current cylinder is the specific cylinder, the calculated learning correction value is stored and updated as it is by the first learning value storage means, and for the other cylinders, only the difference from the learning value of the specific cylinder is calculated. The learning value storage means 2 stores and updates the learning value. The engine speed can be calculated from the signal from the crank angle sensor. Further, the load of the engine can be detected in the form of the filling amount based on the intake air amount signal from the air flow sensor.

(Example) Hereinafter, the Example of this invention is described based on drawing.

FIG. 2 is a schematic diagram of an embodiment in which knocking of the engine is controlled by retarding (that is, retarding) the ignition timing, and an ECU forming a main part thereof.
FIG. 3 is a functional block diagram of (electrical control unit). The ignition plug 7 provided on the upper part of the combustion chamber 9 is connected to the ignition coil 5 via a distributor 6, and the ignition coil 5 is
It is connected to ECU3. The distributor 6 is provided with a crank angle sensor and a cylinder sensor. This crank angle sensor outputs a reference signal to the ECU 3. In FIG. 2, reference numeral 1 denotes an engine body, and a cylinder block 10 thereof is provided with a knock sensor 2 for detecting knocking from vibration of the cylinder block 10. A surge tank is formed in the intake pipe 12 above the intake manifold 11, and an air flow sensor 8 for detecting the intake air amount of the engine is provided in the intake passage immediately above and upstream of the throttle valve 13. . The intake air is taken in from an air cleaner (not shown), and the injector 4
Together with the fuel injected from the combustion chamber 9 via the intake valve 14.
Supplied within. The exhaust gas after combustion flows from the exhaust manifold 16 through the exhaust valve 15 to the exhaust pipe 17, and is exhausted through the catalyst device 18.

The ECU 3 receives the output of the knock sensor 2, the output of the crank angle sensor, and the output of the air flow sensor 8 as input, and receives a signal from the cylinder sensor to perform an operation for each cylinder to output an ignition signal.

In FIG. 3, the ROM contains basic data and a program for controlling the ignition timing. The output of the air flow sensor and the output of the knock sensor are input to the A / D converter through the analog buffer. In addition, the signal from the cylinder sensor is input to the input port through the digital buffer. The CPU exchanges data with the ROM, the A / D converter and the input port, and with the PTM (programmable timer), the free running counter and the learning value storage means. On the other hand, the crank angle sensor generates a pulse-shaped signal for each predetermined crank angle, and the pulse-shaped signal is waveform-shaped and enters INT (interrupt) to take an interrupt timing. The program flow is controlled based on the crank angle signal. The output of the PTM is converted by the output interface into a signal for delivery to the ignition coil.

FIG. 4 is a timing chart showing basic control of ignition timing. This embodiment has four cylinders, and a crank angle signal is output every 60 degrees ATDC (after top dead center). Since it has 4 cylinders, 180 between TDC (top dead center) and TDC
It is degree. When the crank angle signal comes, the calculated energization time TC until ignition is written in PTM. In other words, the value of TC is set in PTM. After TC is set, the counter PTM starts subtraction at the time when the negative edge (falling point) from the crank angle signal comes. As soon as the time information TC is written, the PTM outputs a high signal (the opposite of the ignition signal shown in the figure) and falls low when the counter value reaches 0 after the time TC has elapsed. This high time is the time when the ignition coil is energized, and when the output of the PTM drops to low, the voltage of the ignition coil drops sharply and sparks fly to the ignition coil. By the way, a cylinder signal is input to the input port (see FIG. 3), and when this signal is high, it means that the cylinder is the first cylinder (specific cylinder), so that the time point indicated by the arrow in FIG. This is the ignition timing of one cylinder. If the cylinder signal is low, it means that the cylinders are other cylinders. Therefore, the ignition timing is the ignition timing of the third cylinder, the time is the ignition timing of the fourth cylinder, and the ignition timing is the ignition timing of the second cylinder according to the ignition order. become. Basically, the ignition timing is controlled in this way.

As shown in FIG. 5, the learning map expressing the learning value for each operating region of the engine is 2 bytes for the specific cylinder (first cylinder) and expresses the retard amount with respect to the basic advance angle as it is. For the cylinder No., the amount of delay angle with respect to the learning value of the specific cylinder is represented by 1 byte. The vertical axis of the map is the rotation speed, and the horizontal axis is the filling amount.

FIG. 6, FIG. 7 and FIG. 8 show flow charts for executing such control.

Of these, FIG. 6 shows a background routine that constantly flows. In addition, B _{1 to} B ₁₄ in the figure indicate each step.

After the initialization is performed in B _2, to calculate the engine rotational speed NE from the TDC period. The TDC cycle corresponds to T0 shown in FIG. 4, and is obtained by an interrupt routine described later and is taken in here. In the case of four cylinders, the interval between TDC and TDC is 180 degrees, so the engine speed can be found by looking at the time between these.

Enter the output QA of the air flow sensor with B ₄ , and fill with B ₅
Calculate CE. Since QA is the amount of air taken into the engine per unit time, the filling amount for each cylinder is calculated from this. This corresponds to the load on the engine.

In B _6, based on the charge amount CE and the rotational speed NE, to calculate a basic ignition timing AB from the map. I will call this because the map contains the ignition timing obtained in advance through experiments.

Then, in the B _7, the current operating area to see whether or not the knock zone. Here, if the filling amount CE is 300 CC or more, it is determined to be in the knock zone.

NO, that is, not in the knock zone,
Clear the knocking control flag (B _8), to clear the learning correction value (B _9). Then go back to B ₃ .
That is, the control is performed only by the basic ignition timing in a place other than the knock zone.

If YES, that is, in the knock zone, set the knock control flag (B ₁₀ ), and go to B ₁₁ .

At B ₁₁ , the current learning zone is calculated from the map of the rotational speed NE and the filling amount CE. That is, the row number j and the column number k in the map are calculated to determine which zone j, k the current operating state corresponds to. Then, the learning value L ijk corresponding to the current learning zone j, k is read from the map on the RAM (B ₁₂ ). In other words, the number of the cylinder currently controlled and the map to be selected are determined by i
, And read L ijk as the final learning value.

Then, the learning correction value AL1 for the first cylinder is calculated (B ₁₃ ),
Then calculates the learning correction value of the other cylinders (B _14).
If the current cylinder is the first cylinder, the data is used as it is, but if it is another cylinder, the data AL1jk of the first cylinder and the data L ijk of the cylinder are added.

 The above is the background routine.

Next, the interrupt routine shown in FIGS. 7 and 8 will be described. I _{1 to} I ₂₈ indicate each step.

The interrupt routine is activated once for each ATDC 60 degree signal. The background routine is operating all the time, but when the ATDC signal of 60 degrees arrives, it jumps to the interrupt routine, and when this ends once, it returns to the jumped position. That is the kind of control overall.

This routine starts with the crank angle signal of ATDC 60 degrees shown in Fig. 4 ( ₁₁ ), and the time T when the interrupt occurs is read from the FRC (free running counter) (I ₂ ), for example, the time read this time. Is T1, then T1
From the time T2 read in the previous time and TDC cycle T0
(I ₃ ). Then, in order to calculate the cycle again next time, this time T1 is put into T2. That is, the last interrupt time is updated (I ₄ ).

Next, I ₅ detects whether it is a knock zone.
That is, it is detected whether or not the flag set here is 1.

If it is not in the knock zone (NO), the feedback correction value is cleared (I ₆ ).

If it is the knock zone (YES), the knock intensity IK is input (I ₇ ). Then, whether or not there is knocking (I ₈ ), NO, that is, if there is no knocking, go to I ₉ and correct the feedback correction value obtained so far to the advance side. This is a predetermined value (very small value) from the previous feedback correction value
It is an operation to obtain a new correction value by subtracting.

If knocking (YES), go to I _10, the intensity value corresponding to knocking (CR × IK, however, CR is a constant, IK is knock intensity) was added to the feedback value, the non-knocking I ₁₁ Clear the counter and go to I ₁₂ .

In I ₁₂ , it is checked whether it is a learning condition. In other words, if the knocking is strong, the learning value has to be corrected to the retard side, and it is checked here whether the detected knock intensity is larger than the learning determination knock intensity IKL (constant).

And yes, only when knocking is strong,
Correct the learning correction value of the current cylinder to the retard side (I ₁₃ ).
Specifically, it is an operation of adding a value obtained by adding AF / Bi × 0.1 to the learning value ALi so far to obtain a new learning value. Modifications are made in minute increments so that it is possible to deal with situations where sudden knocking occurs but not later.

In I _12, NO, namely if that is not the learning condition, no updating of the learning value.

When the learning value is retarded in I ₁₃ , the process proceeds to the step of updating the learning value. First, in I ₁₄ , it is checked whether i is 1, that is, whether it is the first cylinder. Is 1
If so (YES), the learning value for the first cylinder is updated (I ₁₅ ).

If i is not 1 (NO), the learning value of the current cylinder is updated (I ₁₆ ). This is an operation of obtaining the difference between the correction value ALi finally obtained for that cylinder and the learning value L1jk of the first cylinder, and finally storing the difference corresponding to this difference.

When the learning value is updated in this way, the final ignition timing is calculated based on this, and at this time, I
_{At 17} , it is determined what cylinder the cylinder at that time is. That is, looking at the cylinder signal coming into the port shown in FIG. 3, if this signal is high, it means the first cylinder, if it is low, it means the other cylinder, and if it is high (YES ), The cylinder number i is initialized at I ₁₈ (i is set to 1), and if low (NO), the cylinder number i is updated at I ₁₉ (i + 1 is added to i + 1).
Here, for example, when i is 2, it means that the current cylinder is the second cylinder in the ignition order, that is, the third cylinder.

After determining the working cylinder in this way, the final ignition timing is calculated by I ₂₀ .

On the other hand, in the knocking region, when the knocking does not occur, the feedback correction value is corrected to the advance side with I _{9 as} described above, and then with I ₂₁ ,
I am looking at whether knocking has occurred in the past.
Specifically, the learning value up to now in that driving state is 0.
If it is larger than that, it is determined that knocking has occurred in the area in the past. If knocking has not occurred in the past (NO), the learning correction value is 0 and there is no update. In the case where knocking is occurring (YES) in the past, the non-knock counter of the current cylinder increment (CW / Oi = CW / Oi + 1) and in I _22, see whether it is the learning conditions in the I _23. In other words, because there is no knocking (I ₈ ) I am here, so I have to correct the learning value to the advance side, but if I correct it immediately, hunting will occur, so knocking will definitely occur. To see if it doesn't happen, look to see if you've been knocked free over 300 times. If knocking does not occur more than 300 times, I ₂₄ is used to correct the learning value of the current cylinder to the advance side, and the non-knock counter of the current cylinder is cleared (I ₂₅ ).

In this way, one is an operation to correct the learning value to the retard side because a large knocking has occurred, and the other is to correct the learning value to the advance side because knocking did not occur over 300 times. The learning value is updated in steps I _{14 to} I ₁₈ based on these operations. Then, to calculate the final ignition timing reaches a step I ₁₇ ~I ₂₀ as described above. The final ignition timing AS is the feedback correction value from the basic ignition timing AB
AF / Bi and learning value ALi are subtracted. If it is not in the knock zone (I ₅ = NO), AF / Bi and ALi are both 0.
Therefore, AS = AB.

Next, I ₂₆ is used to calculate the time to ignition TC.

As is clear from the timing chart of FIG. 4, the crank angle corresponding to time TC starts from 90 degrees and the ignition timing AS (TD
Before C) is subtracted. The time TC can be calculated from this crank angle. The energization time TC calculated in this way is PTM
Set to (I ₂₇ ). As described in the explanation of FIG. 4, the PTM outputs a high signal from the set time, and the time T
When C elapses, the TPM output drops low. This time is the ignition timing. In this way, the PTM is managed on time.

The interrupt routine starts by the crank angle signal, as seen in the timing chart of FIG.
If the cylinder signal is high when the crank angle signal comes, the cylinder number i is 1 and TC calculated here is TC.
Based on the above, ignition timing control of the specific cylinder (first cylinder) is performed. Then, at the next crank angle signal, the interrupt routine starts again. Since the cylinder signal is low at that time, i becomes 2 (third cylinder) from this point. Thereafter, i is updated to 3 (4th cylinder) and 4 (2nd cylinder), and is cleared to 1 by the next high cylinder signal.

By the way, the present invention is not limited to the knocking control as described above, but may be, for example, one in which the air-fuel ratio is controlled for each cylinder in a multi-cylinder engine. Even if the fuel is supplied in the same way, it is unavoidable that the air-fuel ratio varies between the cylinders, so in order to control the air-fuel ratio of each cylinder to, for example, 14.7, each cylinder has its own correction value, Moreover, the manner of holding the correction value can be set as described above. Further, in the above-described embodiment, the learning value is expressed in 2 bytes for a specific cylinder and expressed in 1 byte for other cylinders. Can also be expressed as 1/2 byte.

(Effects of the Invention) Since the present invention is configured in this way, it is possible to reduce the storage capacity without reducing the overall accuracy of the learning control for each cylinder. Therefore, since it is not necessary to add RAM as in the conventional case, the effects of cost reduction, noise resistance improvement, space saving, and weight reduction can be obtained.

[Brief description of drawings]

FIG. 1 is an overall configuration diagram of the present invention, FIG. 2 is a schematic diagram showing an embodiment of the present invention, FIG. 3 is a functional block diagram showing an essential part of the same embodiment, and FIG. A timing chart showing the basic control of the ignition timing, FIG. 5 is a diagram showing a map of the retarded amount learning value in the same embodiment, FIG. 6, and FIG.
FIG. 8 and FIG. 8 are flowcharts for executing the control of the same embodiment. 1: Engine body, 2: Sensor, 3: Electric control device, 7: Distributor, 9: Air flow sensor.

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical indication location F02D 45/00 340 H 376 B F02P 5/15 5/152 5/153

Claims (1)

[Claims] 1. An operating state detecting means for detecting an operating state of an engine, a basic control amount determining means for determining a basic control amount based on an output of the operating state detecting means, and a combustion state for each cylinder of the engine. Combustion detection means for each cylinder, feedback correction amount determination means for determining a feedback correction amount for each cylinder according to a signal from the combustion detection means for each cylinder, and learning value calculation means for calculating a learning value based on the feedback correction amount. , Cylinder identifying means for identifying the current cylinder,
First learning value storage means for receiving a signal from the cylinder identification means and storing a learning value of a specific cylinder; difference calculation means for calculating a difference between a learning value of another cylinder and a learning value of the specific cylinder; Second learning value storage means for storing the output of the difference calculation means as a learning value for the other cylinder, and a final control amount for determining a final control amount for each cylinder from the basic control amount, the feedback correction amount, and the learning value. And a combustion control means for controlling the combustion state of the engine by the output of the final control amount determination means. The learning value of the specific cylinder is stored as it is and the learning value of the other cylinders is the learning value of the specific cylinder. The engine control device is characterized in that the difference is stored as a difference.