JPH0797954A - Knocking detection method of internal combustion engine and ignition timing controller using it - Google Patents

Abstract

PURPOSE:To surely prevent erroneous judgement of knocking by setting a back ground level initial value at the time of entering into a knocking control range, in a ratio type knocking control method using a resonant frequency where knocking control is limited only in a specified operational range. CONSTITUTION:An injector 16 and an ignition coil 13 are respectively controlled by a control unit 9 based on respective detection signals from respective sensors 2, 5, 11, 12, 200 for respectively detecting operational condition of an engine 7. Namely, characteristic component of knocking is extracted and comparison component is calculated so as to judge the knocking based on the detection signal particularly from the combustion condition sensor 200. In this case, the comparison component is calculated only in a specified operational range, and then the calculated value of the comparison component at the time of separation from this range is maintained. On the other hand, the calculated value of the maintained comparison component is compared with a standard value of the comparison component preliminarily determined at the time of reentering into the specified operational range. The larger value can be made the initial value of the comparison component at time of entering.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for detecting knocking of an internal combustion engine and an ignition timing control method using the same, and more particularly to reduction of erroneous detection of knocking due to change in operating condition of the internal combustion engine.

[0002]

2. Description of the Related Art As is well known, knocking is a phenomenon in which gas in a combustion chamber vibrates due to self-ignition of unburned gas in a terminal portion of the combustion chamber, and the vibration is transmitted to an engine body.

Since this knocking causes a loss of energy generated by the engine (a reduction in output), an impact on various parts of the engine, and a reduction in fuel consumption, it is desirable to avoid it as much as possible, and for that reason, knocking occurs. Accurate detection is essential. Under such a demand, for example, as disclosed in JP-A-58-45520, only the resonance frequency component in the range of 5 to 12 kHz is separated from the output signal of the vibration detection sensor by using a band pass filter. Then
A method of detecting the occurrence of knocking depending on whether the integrated value of the output becomes larger than the background level or not, as described in Japanese Patent Laid-Open No. 3-47449, knocking out a plurality of resonance frequency components. There was a method to detect.

By the way, the background level is
For example, as shown in Equations ₁ and 2, the weighted average of the background level BGL ₁ of the previous time and the output signal of the vibration detection sensor f ₁ of this time, or the output f ₂ of the vibration detection sensor of the previous time and the vibration of this time Detection f ₁ The output of the sensor is obtained by weighted averaging.

[0005]

## EQU1 ## BGL = kBGL ₁ + (1-k) f ₁ (Equation 1)

[0006]

[Number 2] _{BGL = kf 1 + (1-} k) f 2 ... ( Equation 2) k: constant Therefore, the techniques described above, in transient acceleration or the like,
As shown in FIG. 9, there is a background level following delay, and during the transition, the calculated background level value is smaller than the target background level value. There is a problem that the possibility of false detection of knocking due to an increase in the N ratio increases.

As a technique for solving this problem, for example, as disclosed in Japanese Patent Application Laid-Open No. 3-124941, the background level is constantly learned and stored for each operating region, and the operating region in which knocking is detected is detected. A method has been proposed in which knocking determination is performed by using the learned and stored background level instead of the calculated background level at the time of rush or transition.

[0008]

However, in the above-mentioned conventional technique, since the learned background level is used as it is for knocking determination, the learned value is lower than the background level required at the time of inrush. If the value has been learned, there is a risk of erroneous knock determination.

[0009]

According to the present invention, the above-mentioned problems are provided in the vicinity of a combustion chamber of an internal combustion engine, the combustion state detecting means for detecting the combustion state of the internal combustion engine, and the combustion state detecting means. In the method of extracting a characteristic component related to knocking from the state signal of, performing a calculation of a comparison component serving as a reference for knocking determination based on the state signal, and performing knocking determination based on the characteristic component and the comparison component. , The calculation of the comparative component only in a predetermined specific operating region, holds the calculated value of the comparative component when leaving the specific operating region, at the time of reentry of the specific operating region, of the held comparative component A value is compared with a comparison component reference value which is a reference value of a comparison component of the predetermined specific operating region, and the larger value is set as the initial value of the comparison component at the time of the rush. To be solved by performing the calculation of said comparison components based on the initial value, and the determination of knocking.

[0010]

According to the present invention, the learned and held background level is compared with the reference background level, and the larger background level is set in the specific operation area (hereinafter referred to as knock control area). Since the background level is set to the initial value at the time of entry, the initial value can be set to an appropriate value, and false determination of knocking can be reduced. Further, since the reference value is learned and a limit value is set for the learned value, it is possible to prevent the reference value from becoming an abnormal value.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described in detail below with reference to the drawings.

First, the principle of determining whether knocking occurs in the present invention will be described. The vibration of the engine contains many vibration components. For example, there are vibration components due to piston friction, crankshaft rotation, valve operation, and the like. Further, these vibration components change depending on the engine state.

When knocking occurs in the engine, vibration unique to knocking occurs. Whether or not knocking has occurred is determined by separating the vibration characteristic of knocking from the overall vibration of the engine detected by the vibration sensor.

FIG. 4 is a diagram showing the analysis result of the frequency component of the output of the vibration sensor when knocking does not occur. On the other hand, FIG. 5 is a diagram showing the analysis result of the frequency component of the output of the vibration sensor when knocking occurs.

As can be seen by comparing FIGS. 4 and 5, each characteristic component, that is, the resonance frequency component, becomes larger when knocking occurs than when knocking does not occur. Can understand.

Next, with reference to FIGS. 6 and 7, a description will be given of the determination of the occurrence of knocking using the knocking determination index. In order to explain the principle operation, the resonance frequency f ₁₀
(Eg 6.3 KHz) and f ₀₁ (eg 13.0 KHz)
This will be described using the frequency components of. However, the present invention is not limited to this, and the presence or absence of knocking can be determined using any two or more resonance frequency components.

The vibration sensor combines and detects vibration including knocking and background vibration.
Therefore, knocking determination index I is determined include vibration I _k due to the occurrence of background vibration I _b and knocking when index I _b becomes defined in the background vibration, the knocking occurs when a knocking has not occurred It is the index I that can be obtained.

The above knocking determination index I is mathematically expressed by using the main resonance frequency components as follows.

I = ω ₁₀ P (f ₁₀ ) + ω ₂₀ P (f ₂₀ ) + ω ₀₁ P (f ₀₁ ) + ω ₃₀ P (f ₃₀ ) + ω ₁₁ P (f ₁₁ ) where ω is determined by the engine speed Take a real value. It can also take a binary value of 1 or 0. P is the vibration intensity (power spectrum) of each resonance frequency component.

As shown in FIG. 6, the knock determination index I _b indicated by the resonance frequency component of the background vibration and the index I _k indicated by the resonance frequency component of the vibration due to knocking are different in direction and magnitude. . This is because, as is apparent from a human hearing test, the engine sound when there is no knock is distinct from the engine sound when there is knock, and the tone color is different depending on whether there is knock.

When the vibration due to the occurrence of knocking is added to the background vibration, the knock determination index I based on the f ₀₁ and f ₁₀ components included in the vibration sensor enters a region below the knock determination threshold I ₀₁ in the case of FIG. Further, in FIG. 7, it is possible to determine whether knocking has occurred by going outside the threshold value I ₀₂ .

In the present specification, not only the five terms on the right side of the equation (1), but also a combination of a plurality of resonance frequency components included in the output of the vibration sensor is defined as a knock index. .

As described above, when the knock determination index is used, the configuration of the specific frequency component due to the occurrence of knocking with respect to the background vibration is taken into consideration, so that the presence or absence of knocking can be determined even if the background vibration becomes large. .

The knocking detection method, which is the core of the present invention, will be described in detail below.

FIG. 8 shows the vibration intensity (power spectrum) for each frequency. A solid line shows the case where knocking occurs, and a broken line shows the case where knocking does not occur. It can be seen that the vibration intensity is high in each resonance frequency band.

In the present invention, the above-described knock index is obtained by the ratio of vibration intensity (S / N ratio) with and without knocking, but the vibration intensity (background level) without knocking is detected for each detection frequency. It is created by averaging processing based on the knock frequency components detected in the past.

Therefore, when there is a sudden change in the vibration intensity without knock due to an increase in mechanical noise due to a sudden change in engine speed such as a transient operation state or an increase in combustion vibration due to a sudden change in engine load, the background level has the latest knock frequency component value. However, as a result, there is a risk of erroneous determination of knocking shown in FIG.

In many cases, knock control is defined only in a specific region determined by the engine speed and engine load. In this case, the background level smoothing process is performed only in the specific knock control region and in the region. There are two types of processing methods, i.e., a method that is performed regardless of the above. Both of these will be described with reference to FIG.

In the former method, the background level when the knock control area has left the previous time is set as it is as the initial value of the background level when the knock control area is re-entered. If the initial value of the background level at this time is abnormally high, the background level during a transition quickly converges to the target value, but on the other hand, knocking is difficult to detect until it converges to the target value. Become. In the latter method, the background level at the time of entry into the knock control area is close to the target value, so knocking can be detected with high accuracy, but trackability deteriorates during transients and the background level quickly converges to the target value. become unable.

As described above, since the two have the contradictory characteristics, the smoothing of the background level is performed in such a manner that the respective characteristics of the two are utilized and the respective defects are offset. Processing must be done.

Therefore, in the present invention, the smoothing process of the background level is limited to the knock control region, and the background level is converged from an appropriately high value to improve the tracking capability of the background level even during the transient operation, thereby causing an erroneous determination. It is possible to suppress the occurrence of knocking and secure the knocking controllability at the time of entering the knocking control area.

First, the operation of the process of determining whether or not knocking has occurred by the CPU will be described with reference to FIG.

In FIG. 1, this flow chart is executed every explosion cycle and is activated by interrupting the CPU.

In step 140, it is determined whether or not the current operating state is within the knock control region. If it is within the region, the process proceeds to step 141 below, and if it is outside the region, knock control is not performed. It should be noted that this knock control area determination is performed in step 1.
It may be set in the preceding stage such as 02, 105, 106.

At step 141, it is judged whether or not the vehicle was in the knock control area at the time of the previous explosion cycle, that is, when the knock area is entered, and it is judged that it was in the knock area last time and was continuously in the knock area. If step 1
When it is determined that the previous time is outside the knock control area and the knock area is entered, the routine proceeds to step 01, and at step 142, the initial value BGLINTi preset in the memory is set in BGLi, and the routine proceeds to step 101. In step 101, the output signal from the vibration sensor is converted by the A / D converter and the A / D converted value is fetched.

Next, at step 102, the A / D converted signal of the vibration sensor is subjected to frequency analysis. This frequency analysis is performed by a method such as fast Fourier transform or Walsh Fourier transform.

Thereafter, among the signals subjected to frequency analysis in step 103, n frequency bands including resonance frequencies are selected. In this embodiment, eight resonance frequencies are selected.

When the frequency is selected in step 103,
Next, at step 104, the S / N ratio representing the vibration intensity is obtained for each selected frequency.

That is, a plurality of selected frequencies (f1 ... f
n) and the background level (BGL1 ... BGLn) of the frequency corresponding thereto are calculated, and the S / N for each frequency is calculated.
The ratio SLi = fi / BGLi is determined.

Therefore, in this embodiment, SL1 = f1 / B
GL1, ..., SLi = fn / BGLn is obtained.

Step 132 is a counter, which is incremented by one each time the S / N ratio of one selected frequency is calculated once. In step 133, it is determined whether or not the value of the counter value i is equal to the selected number n, for example, 8, and if n = i,
Proceed to step 105, and if n> i, step 129
Then, the above-mentioned judgment is repeated until all S / N ratios of the selected frequency are calculated.

Next, in step 105, among the selected frequencies, m, 5 in this embodiment, are extracted in descending order of S / N ratio to obtain the knock strength. The formula for obtaining this knock intensity is, for example,

[0043]

[Equation 3]

It is obtained by adding the S / N ratio as represented by.

When the knock intensity is obtained, the predetermined value for knock determination is compared with the knock intensity obtained in step 105 in step 106, and knocking occurs when it is determined that the knock intensity obtained in step 105 is high. As a result, knocking is detected in step 107.

Thereafter, at step 108, the knock flag indicating the occurrence of knocking is set to "1". This knock flag is used in the ignition control task which is activated separately.

On the other hand, if it is determined in step 106 that the knock intensity is smaller than the predetermined value, it is determined that knocking has not occurred, and in step 109 each background level BGL.
It is determined whether i is greater than a predetermined limit value, here the lower limit limiter BGLMTi. In this embodiment, BGL1
… BGL8 is compared with BGLMT1 …… BGLMT8.

When it is determined in step 109 that the background level is higher than the lower limit limiter BGLMTi, that is, when it is determined that the background level is normal, the background level BGLi is updated in step 110.

The updating of the background level BGLi is obtained by filtering the vibration intensity of the selected frequency. Specifically, for each selected frequency, BGL
i = BGLi * (1- [alpha]) + fi * [alpha].

On the contrary, if it is judged at step 109 that the background level BGLi is smaller than the lower limit limiter BGLMTi, that is, if it is judged as an abnormal background level, the limiter value is set and used as BGLi at the next step 104.

A lower limit limiter fLMTi for the selected frequency fi is provided before the step 104 to suppress the input of abnormal values.

In step 129, it is determined whether or not the intensity is greater than or equal to the predetermined lower limit limiter fLMTi for each frequency fi. If it is judged to be greater than or equal to this judgment, the routine proceeds to step 104, where normal S /
Perform N-ratio calculation.

On the other hand, if it is determined in step 129 that the intensity is smaller than the predetermined lower limit limiter fLMTi for each frequency fi, the routine proceeds to step 130, where the S / N ratio between the intensity of the frequency fi and the background level is set to 1. . That is,
A process is performed that determines that knocking has not occurred for the corresponding frequency.

Next, at step 131, the intensity of the corresponding frequency is replaced with a predetermined lower limit limiter fLMTi, and the routine proceeds to step 105, where knock intensity is calculated.

This process solves the problem of abnormal increase in S / N ratio due to quantization error that occurs when the intensity of each frequency is abnormally reduced by setting the lower limit limiter to the frequency intensity.

Here, the lower limit limiter fLMTi of each frequency fi in step 129 and the lower limit limiter BGLMTi of the background level in step 109 may be used together or only one of them may be adopted.

Next, at step 112, the knock flag is set to "0".

The knock detection routine is completed by the above processing, but the knock flag set in this routine is used in the ignition control task. Although the lower limit limiter is shared with the threshold for judging the background level, it may be provided separately.

The knock generation signal obtained as described above is used in the ignition control task described below, and therefore its explanation will be given.

FIG. 10 is a system configuration diagram of the ignition device. Air enters from the inlet of the air cleaner 1, passes through the dust 3, the throttle body 5 having a throttle valve 5, the intake cylinder 6, and is sucked into the cylinder of the engine 7. The intake air amount is detected by the hot-wire air flow meter provided in the duct 3, and the detection signal is input to the control unit 9.

On the other hand, fuel is injected from a fuel tank (not shown) through the injector 16, mixed with intake air in the intake passage and supplied into the cylinder of the engine 7.
The air-fuel mixture is compressed by the engine 7, ignited by the ignition flag 15, and discharged from the exhaust pipe 8 after the explosion. Exhaust pipe 8
An exhaust sensor 11 is provided in the control unit 9 and the detection signal is input to the control unit 9.

The high voltage generated in the ignition coil 13 is distributed to each cylinder by the distributor 14 and supplied to the ignition plug 15. The rotation state of the engine is detected by the crank angle sensor 12, and the crank angle sensor 12 outputs a Ref signal indicating an absolute position for each rotation and a P _OS signal indicating a position moved by a predetermined angle from the absolute position. The Ref signal and the P _OS signal are input to the control unit 9. A vibration sensor 151 for detecting vibration is attached to the engine 7, and the detection signal is input to the control unit 9.

The control unit 9 calculates the fuel supply amount and the ignition timing based on the signals from the respective sensors, and outputs the control signals to the injector 16 and the ignition coil 13. FIG. 11 is a diagram showing details of the control unit 9. The control unit 9 is a CPU 20, A / D
Converter 21, ROM 22, input I / O 23, RAM 2
4, DPRAM25, control block 34 composed of output I / O 26 and bus 37, CPU 29, port 2
7, timing circuit 28, A / D converter 30, ROM 3
1, a RAM 32, a clock 33, an operational circuit 38, and a knock detection block 35 including a bus 36. Here, CPU20, CPU
DPRA is a dual port RAM for exchanging 29 data
It is done through M25.

The intake air amount Q _a detected by the hot wire type flow meter 2 is converted into a digital value by the A / D converter 21 and taken into the CPU 20. The Ref signal and the P _OS signal detected by the crank angle sensor 12 are taken into the CPU 20 through the input I / O 23. C
The PU 20 performs arithmetic processing according to the program stored in the ROM 22, and the arithmetic result is the fuel injection time signal Ti and the ignition timing signal θig that mean the fuel injection amount from the output I / O 26.
It is transmitted to each actuator as n. The RAM 24 holds necessary data during arithmetic processing.

On the other hand, in the timing circuit 28, the operation circuit 37 is a TDC indicating the top dead center.
When the signal is generated, the CPU 20 divides the periodic signal generated by the clock 33 according to the content input to the port 27 to generate a sampling signal. When the sampling signal is generated, the A / D converter 30 converts the output signal P of the vibration sensor 151 into a digital value.

As the vibration sensor for detecting knocking, the conventional one resonates at around 13 KHz, but in the present embodiment, one that resonates at 18 KHz or more is used to obtain a resonance frequency component of at least 18 to 10 KHz. ing.

The CPU 29 outputs the digital value sampled according to the program held in the ROM 31 to the RA
In addition to being stored in M32, frequency analysis is performed based on the stored data based on the flowchart as shown in FIG. 1 to determine whether knocking has occurred. The determination result of whether or not knocking has occurred is sent to the CPU 20 via the DPRAM 25.
Be transmitted to.

Next, the operation of calculating the ignition timing by the CPU 20 will be described with reference to the flowchart of FIG. The operation of this flow chart is activated every fixed time period, for example, every 10 msec. In step 201, the engine speed N and the intake air amount Q are read from a predetermined register set in the RAM 24. In step 202, the intake air amount Q / N per unit rotational speed is calculated, the fuel injection time width Ti is obtained from Q / N, and the basic ignition timing map stored in the ROM 22 for fuel supply is used to perform basic ignition. Timing θ
ask for base. In step 203, the presence or absence of knocking is determined based on the content of the knock flag (knockflag) determined by the flowchart shown in FIG. If knocking has occurred, at step 213 the ignition timing θad
Subtract a predetermined amount of retardation Δθret from v. The ignition timing is retarded by this subtraction. At step 214, the ignition timing retarded due to knocking is compared with a predetermined rotation, for example, 50 (step 2
In 05), the base to be recovered is decided. The count data A is initialized and the process proceeds to step 208.

On the other hand, if knocking has not occurred in step 203, in step 204 the count data A
Count up by one. The count data A indicates the ignition timing θadv retarded by the occurrence of knocking and the advance amount Δθa.
It is used to determine when it is time to recover dv. In step 205, it is determined whether the count data A has become equal to the predetermined value 50. Since the flow shown in FIG. 16 is started every 10 msec, when the count data A becomes equal to 50, it means that 0.5 seconds has elapsed since the count data A was initialized, and 0.5 seconds has passed. It is recovered every time. If the count data A is not equal to 50 in step 205, step 20
Go to 6. In step 206, a predetermined advance amount Δθadv is added to the retard value θadv. The ignition timing is recovered by this addition.

Next, at step 208, the basic ignition timing θba
The ignition timing θign is calculated by adding the ignition timing θadv obtained as described above to se. In step 209, the engine speed N and the intake air amount Q / unit speed Q /
The maximum advance value θres is calculated according to N. Maximum advance value θ
res is done by reading from the maximum advance value map stored in the ROM 31. In step 210, it is determined whether the ignition timing θign exceeds the maximum advance value θres. If not exceeded, the process proceeds to step 211. If the maximum advance angle value θres is exceeded, the advance angle is too large, so in step 211 the maximum advance angle value θres is set as the ignition aging θign.

Finally, after the ignition timing θign is set,
In step 212, the delay time td, the number of sampling points ns, and the frequency division ratio ts are set in the port 27 according to the engine state.
Output to.

The frequency division ratio ts determines the sampling cycle of the digital value of the output of the vibration sensor, and the sampling point number ns determines the sampling point.

By thus detecting knocking from a plurality of resonance frequency components and controlling the ignition timing, knocking of the engine can be avoided.

By the way, the output from the vibration sensor attached to the cylinder block has a characteristic of increasing with an increase in engine speed and load.

This is due to mechanical noise such as piston sludge in the engine and change in combustion mode.

Although it has been shown that the knock index is created from the knock frequency component and the background level in the above-mentioned knock index creation, the background level is obtained by averaging the knock frequency components and has a certain time constant. Therefore, in the transient operation state, it always lags behind the new value of the knock frequency component.

This means that the denominator side is apparently small in the ratio method of detection for each frequency as the premise of the present invention, and the knock index after calculation is large.

Since knock determination is performed by comparing the knock index with the knock determination threshold value, the clearances of both are extremely small in a transient state, and as a result, the knock detection means is very sensitive. In the worst case, the relationship between the two is reversed, resulting in an erroneous knocking determination that knocking occurs even though no knocking occurs (FIG. 9 described above).

As described above, there are two types of background level smoothing processes, that is, only in the specific knock control area and in the case where the background knock level is independent of the area. In the present invention, when the knock control is defined only in the specific region determined by the engine speed and the engine load, the background level smoothing process is limited to the knock control region. This is because by setting the background level update initial value to the optimum value when entering the knock region, the background level is made to converge to an appropriate value and higher than the current value, and the background level follows even during transient operation. This is because it is possible to improve the controllability and suppress the occurrence of erroneous determination, and it is possible to suppress the sensitivity deterioration of the knock controllability at the time of entering the knock control region.

A detailed example of the initial value setting at the time of entering the knock control area will be described below.

As one example thereof, as shown in step 142 of FIG. 1, the background level reference value preset in the memory is held as BGLINTi, and this BGBLINTi is entered when the knock control area is entered. BGL
i is set as an initial value, and the S / N ratio is calculated in step 104 using the initial value, knocking determination in steps 105 to 106, and BGL in step 110.
i is averaged.

This BGLINTi may be set for each engine speed, load, cylinder, and detection frequency, as shown in FIG.

Next, a second embodiment will be described with reference to FIGS. 13 to 14.

In step 301 of FIG. 13, it is determined whether the knock control area has left. If the knock control area has left, in step 302 the final value BGLENDi of BGLi in the knock control area is held in the memory. This step 301 to step 302
May be provided either before or after step 140 in FIG. Further, instead of step 142 in FIG. 1, in step 150 in FIG. 14, the BGLENDi is set to the initial value of BGLi. The subsequent processing is the same as that described in the first embodiment.

According to the second embodiment, it is possible to reduce the matching man-hour when setting BGLINTi described in the first embodiment. Further, the front BGLENDi may be divided into regions by engine speed, load, cylinder, and detection frequency and stored in the memory, as shown in FIG.

Next, a third embodiment will be described with reference to FIG.

Instead of step 142 of FIG. 1, in step 151 of FIG. 15, the detection frequency component (knock frequency component fi) at the time of entry into the knock control region is set as an initial value.

According to the third embodiment, it is possible to contribute to the saving of the memory capacity, and it is effective when a microcomputer system having a small memory capacity is used.

Next, a fourth embodiment will be described with reference to FIG. The fourth embodiment is a further development by combining the first embodiment and the second embodiment.

Instead of the step 142 of FIG. 1, the BGLINTi and the BGLENDi are compared in step 152 of FIG. 16, and the larger one is set as the initial value of BGLi at the time of entry into the knock control area.

Next, a fifth embodiment will be described with reference to FIG. The fifth embodiment is a further development by combining the first embodiment and the second embodiment.

Instead of step 142 of FIG. 1, the BGLINTi and the knock frequency component fi are compared in step 154 of FIG. 17, and the larger one is the BG at the time of entering the knock control region.
Set as the initial value of Li.

According to the above-described fourth and fifth embodiments, the initial value of the background level when the knock control area enters can be set to an appropriate value. With caution, the value can be set near the target value of the background level shown in FIG.

Further, the background level reference value may be learned, and by learning, it is possible to eliminate deterioration of the engine over time and variations in the manufacturing process. This method will be described with reference to FIGS.

In step 140 of FIG. 20, it is determined whether or not it is within the knock control area. If it is determined to be within the knock control region, the background level BGLi calculated in step 156 and the background level reference value BGLI
Compare with NTi. If the BGLi is larger than the previously stored BGLINTi in step 157, the BGLi is replaced by the BGLi.
Holds Li and the next background level reference value BGLI
NTi. Also, instead of step 157 in FIG. 20,
As shown in step 158 of FIG. 21, BGLINTi held last time and BGLi obtained this time are averaged, that is,

[0096]

[Equation 4]

As a reference value BGLINTi for the next background level may be used.

20 and 21 may be started by a routine different from that of FIG. 1, or may be inserted between steps 140 and 141 of FIG. 1, for example. In that case, step 157 or step 158 and step 141 must be connected.

Further, as shown in FIG. 22, the BGLINTi
It is also possible to set a limit value for the learning value of and prevent learning of an abnormal value. For example, a learned BGLI with preset limits
NTi is compared with the limit value, and if the BGLINTi exceeds the limit value, the limit value is held as a background level reference value BGLINTi.

The learning value of BGLINTi described above is different from the memory in which BGLINTi was originally stored (for example, RAM
Etc.), or the learning value or the limit value may be divided into areas as shown in FIG. 16 and held or set.

Further, at the time of correcting the gain of the knock sensor output signal, and although not described, it is preferable to prohibit the update of the background learning value during the specific period after the gain switching. The method will be described with reference to FIG.

In step 140 of FIG. 23, it is determined whether or not it is the knock control area. If it is within the knock control region, it is determined in step 160 whether or not the gain is being switched. If it is not the time of switching, the BGLINTi is learned in step 161.
If there is a switch, learning of BGLINTi is prohibited.

The processing in FIG. 23 may be started in a routine different from that in FIG. 1 or may be inserted between steps 140 and 141 in FIG.

By this method, the erroneous learning factor of the learning value of BGLINTi can be eliminated, and the learning of the abnormal value can be prevented beforehand.

[0105]

According to the present invention, it is possible to prevent erroneous determination of knocking during transition, improve knock detection accuracy regardless of whether engine operation is steady or transient, and optimize ignition timing in all operating conditions. Control becomes possible and it can contribute to the improvement of engine output, fuel consumption, and emission.

[Brief description of drawings]

FIG. 1 is a flowchart showing an embodiment of the present invention.

FIG. 2 is a flowchart illustrating a method for calculating knock intensity.

FIG. 3 is a flowchart illustrating a knock determination method.

FIG. 4 is a vibration waveform diagram when knock does not occur.

FIG. 5 is a vibration waveform diagram when a knock occurs.

FIG. 6 is a diagram regarding a spectrum intensity of knock.

FIG. 7 is a diagram relating to the spectrum intensity of knock.

FIG. 8 is a diagram showing vibration intensity when a knock occurs and when it does not occur.

FIG. 9 is a schematic diagram of a knock misjudgment during a transition.

FIG. 10 is a system configuration diagram.

FIG. 11 is a configuration diagram of a control device.

FIG. 12 is a flowchart showing an ignition calculation.

FIG. 13 is a flowchart for holding BGLi when leaving the knock region.

FIG. 14 is a flowchart showing an example of initial value setting at the time of entry into a knock area.

FIG. 15 is a flowchart showing an example of initial value setting at the time of entry into a knock area.

FIG. 16 is a flowchart showing an example of initial value setting at the time of entry into a knock area.

FIG. 17 is a flowchart showing an example of initial value setting when entering a knock area.

FIG. 18: BGLINTi area division memory storage area.

FIG. 19: Area storage memory for BGLENDi.

FIG. 20 is a flowchart showing an example of initial value learning at the time of entry into a knock region.

FIG. 21 is a flowchart showing an example of initial value learning at the time of entry into a knock region.

FIG. 22 is a diagram showing a limiter for the background level learning value.

FIG. 23 is a flowchart showing prohibition of background level learning during gain switching.

[Explanation of Codes] 1 ... Air cleaner, 2 ... Heat ray type air flow meter, 3 ... Duct, 5 ... Throttle sensor, 6 ... Intake pipe, 7 ... Engine, 8 ... Exhaust pipe, 9 ... Control unit, 11 ... Air-fuel ratio Air-fuel ratio sensor to measure, 12 ... crank angle sensor, 1
3 ... Ignition coil, 14 ... Distributor, 15 ... Ignition flag, 16 ... Fuel injection valve, 200 ... Combustion state sensor.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Masakatsu Fujishita 2520 Takaba, Katsuta-shi, Ibaraki Prefectural Automotive Equipment Division, Hitachi, Ltd. (72) Masami Shida 2520 Takata, Katsuta-shi, Ibaraki Hitachi, Ltd. Automotive Equipment Division (72) Inventor Takayuki Ebisawa 2520 Takaba, Takata, Ibaraki Prefecture Hitachi Ltd. Automotive Equipment Division (72) Inventor Seiichi Kawasaki 2477 Kashima Yatsu Kashima Yatsu, Katsuta, Ibaraki Prefecture 3 Within Hitachi Automotive Engineering Co., Ltd.

Claims (7)

[Claims] 1. A combustion state detecting means for detecting a combustion state of an internal combustion engine is provided in the vicinity of a combustion chamber of the internal combustion engine, wherein a characteristic component relating to knocking is extracted from a state signal of the combustion state detecting means, and the state signal is extracted. On the basis of the above, a method of calculating a comparison component serving as a reference for knocking determination, and a method of determining knocking based on the characteristic component and the comparison component, the calculation of the comparison component is performed only in a predetermined specific operation region. , Holding the calculated value of the comparison component at the time of leaving the specific operation region, and at the time of reentry of the specific operation region, the value of the held comparison component and a predetermined reference of the comparison component of the specific operation region The reference value of the comparative component is compared, and the larger value is used as the initial value of the comparative component at the time of the inrush, and the calculation of the comparative component is performed based on the initial value. Knocking detecting method for an internal combustion engine, characterized in that for determining knocking. 2. A combustion state detecting means for detecting a combustion state of the internal combustion engine is provided in the vicinity of a combustion chamber of the internal combustion engine, and a characteristic component relating to knocking is extracted from a state signal of the combustion state detecting means to obtain the state signal. On the basis of the above, a method of calculating a comparison component serving as a reference for knocking determination, and a method of determining knocking based on the characteristic component and the comparison component, if a predetermined specific operating region is entered, The characteristic component extracted at the time of inrush and a comparative component reference value serving as a reference of the comparative component in a predetermined specific operating region are compared, and the larger value is set as the initial value of the comparative component at the time of inrush. A method for detecting knocking in an internal combustion engine, comprising: calculating the comparative component and determining knocking based on the initial value. 3. A combustion state detecting means for detecting a combustion state of the internal combustion engine is provided in the vicinity of a combustion chamber of the internal combustion engine, wherein a plurality of characteristic components relating to knocking are extracted from a state signal of the combustion state detecting means, In a method of calculating a plurality of comparison components serving as a criterion for knocking determination based on a plurality of feature components and performing knocking determination based on the plurality of feature components and the plurality of comparison components, a predetermined identification Only in the operating region, the calculation of the plurality of comparative components is performed, each of the calculated values of the plurality of comparative components at the time of leaving the specific operating region is held, and when the re-entry of the specific operating region is performed,
The values of the held plurality of comparison components and the plurality of comparison component reference values, which are predetermined for each of the feature components and serve as the reference of the comparison components of the specific operation region, are compared for each of the feature components, and the larger one Is used as the initial value of the comparison component for each characteristic component at the time of the inrush, and the knocking detection method for the internal combustion engine is characterized in that the comparison component is calculated and knocking is determined based on the initial value. . 4. A combustion state detecting means for detecting a combustion state of the internal combustion engine is provided near a combustion chamber of the internal combustion engine, and a plurality of characteristic components relating to knocking are extracted from a state signal of the combustion state detecting means. Based on the plurality of characteristic components,
Performing a calculation of a plurality of comparative components that is a criterion for knocking determination, in the method of performing knocking determination based on the plurality of characteristic components and the plurality of comparison components, if a predetermined operating region is entered, A comparison component reference value serving as a reference of a comparison component of a predetermined operation region for each of the plurality of feature components, and the plurality of feature components are compared for each of the feature components, and the larger value is the feature at the time of the inrush. A knocking detection method, wherein an initial value of a comparison component for each component is set, and the calculation of the comparison component and determination of knocking are performed based on the initial value. 5. The knocking detection method for an internal combustion engine according to any one of claims 1 to 4, wherein the comparative component reference value is learned only in the specific operating region, and the learned value is classified by operating region, A knocking detection device for an internal combustion engine, which is stored for each cylinder, for each one of the characteristic components, or for each region including all of them. 6. The method for detecting knocking of an internal combustion engine according to claim 5, wherein the comparison component reference value is learned if the calculated value of the comparison component obtained this time is larger than the previously stored content. A method for detecting knocking in an internal combustion engine, characterized in that the calculated value is stored and updated. 7. The method for detecting knocking of an internal combustion engine according to claim 5, wherein a limit value is set for the learning value of the comparative component reference value, and the learning value of the comparative component reference value is the learning value. If the limit value is exceeded, the limit value is stored and updated instead of the learned value.