JPH0635939B2 - Knocking detection device for internal combustion engine - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a knocking detection device for an internal combustion engine, and more particularly to a knocking detection device that can accurately detect knocking occurrence from a detection signal of engine vibration.

<Prior Art> When knocking at a predetermined level or higher occurs in an internal combustion engine, not only the output is reduced, but also shock is absorbed.
There is an ignition timing control device that detects knocking and corrects the ignition timing so as to avoid knocking quickly because it adversely affects the exhaust valve and the piston (Japanese Patent Laid-Open No. 58-105036). (See gazette, etc.).

Knocking detection for correcting ignition timing due to occurrence of such knocking is performed as follows.

That is, as shown in FIG. 8, the knock sensor 11 that outputs a detection signal according to the vibration level by the piezoelectric element is installed in the engine 12
Mounted on a cylinder block, etc., the detection signal from the knock sensor 11 is input to the bandpass filter 13 to pass only the signal near the center frequency peculiar to knocking, and half-wave rectification is performed, and then the predetermined value is determined by the integrator 14. In the integration section (for example, ATDC 10 ° -60 °), and the peak value of the integrated value in the integrator 14 (sum of specific frequency component intensities in the integration section) is A / D converted by the A / D converter 15. Input to the microcomputer 16. The microcomputer 16 determines whether knocking occurs based on the difference between the peak of the integrated value when knocking occurs and the peak of the integrated value when knocking does not occur (mechanical vibration level).

<Problems to be Solved by the Invention> Incidentally, the vibration waveform of knocking has a tendency that the vibration level locally increases in a predetermined section.

Therefore, as described above, in the configuration in which the occurrence of knocking is determined by the peak level of the integrated value, that is, the sum of the intensities of the specific frequency components in the integration section, the mechanical vibration level increases on average in the detection section due to factors other than knocking. Even in such a case, there is a possibility that the knocking may be erroneously detected.

In particular, depending on the conditions such as the shape of the engine and the mounting position of the knock sensor, it may not be possible to sufficiently secure the difference in the total sum in a region where the signal level is small.In this case, there is a clear difference in the total sum depending on whether knocking occurs. Since it does not occur, the probability of false detection of knocking increases.

The present invention has been made in view of the above problems, and provides a knocking detection device capable of accurately detecting knocking even when the signal level from the knocking sensor is small, by a knocking detection method suitable for knocking characteristics. With the goal.

<Means for Solving the Problems> Therefore, the knocking detection device for an internal combustion engine according to the present invention is configured as shown in FIG.

In FIG. 1, a vibration sensor is attached to an engine body to detect engine vibration, and a specific frequency component extracting means extracts a specific frequency component from a detection signal of the vibration sensor.

The intensity sampling means samples the intensity of the extracted specific frequency component within a predetermined section for each predetermined period.

The maximum level calculating means calculates the sum of the maximum values of the intensities sampled in a predetermined section of each specific frequency component as the maximum level.

The average level calculating means calculates the sum of average values of intensities sampled in a predetermined section of each specific frequency component as an average level.

The knocking determination means compares the difference between the maximum level calculated by the maximum level calculation means and the average level calculated by the average level calculation means with the background level to determine the presence or absence of knocking.

The background level calculation means, when the knocking determination means determines that there is no knocking, the difference between the maximum level calculated by the maximum level calculation means and the average level calculated by the average level calculation means,
The weighted average value is calculated as the background level.

Further, each of the specific frequency components is provided with a frequency contribution rate storage unit that stores a frequency contribution rate that contributes to knocking of the specific frequency component, and the maximum level calculation unit and the average level calculation unit are the maximum value and the average value of the specific frequency component. In addition, it is also possible to adopt a configuration in which the sum of products obtained by multiplying the frequency contribution rates of the corresponding frequency components is calculated as the maximum value level and the average value level.

Further, as shown by a chain line in FIG. 1, a frequency contribution rate storage unit that stores a frequency contribution rate that contributes to knocking of the specific frequency component for each specific frequency component is provided, and the maximum level calculation unit and the average level calculation unit are The maximum value and the average value of the specific frequency component may be calculated as the maximum level and the average level by summing the values corrected by the frequency contribution rate of the corresponding frequency component.

Further, as indicated by a dotted line in FIG. 1, an intensity change detecting means for detecting an intensity change for each specific frequency component, an intensity change of the specific frequency component detected by the intensity change detecting means, and a predetermined normative change characteristic. And an intensity change correction coefficient calculating means for calculating the intensity change correction coefficient, and the average level calculated by the average level calculating means or the maximum level calculated by the maximum level calculating means is the intensity change correction coefficient. You may comprise so that it may correct | amend and obtain | require.

Further, the weighting coefficient used in the background level calculation means for calculating the background level may be set as a function of the intensity correction coefficient calculated by the intensity change correction coefficient calculation means.

<Operation> According to such a configuration, since knocking detection is performed based on the difference between the maximum level and the average level of the intensity within a predetermined section in which the intensity is sampled at each predetermined period,
It is possible to perform highly accurate knocking detection that matches the knocking vibration characteristics.

Further, if the configuration is such that the maximum level and the average level of the intensity are corrected using the frequency contribution rate, the intensity of the frequency that greatly contributes to knocking can be reflected in the knocking detection at a higher rate, so that the knocking detection can be performed. The accuracy is improved.

Further, if the intensity change correction coefficient is used to correct the average level or the maximum level by the intensity change correction coefficient,
If the vibration characteristic estimated by the intensity change is close to the knocking vibration characteristic, by correcting the average level or the maximum level in the direction in which it is easy to determine that knocking occurs,
The knocking detection accuracy is further improved, and if the weighting coefficient used for the background level calculation is also corrected by the intensity change correction coefficient, the knocking detection accuracy is further improved.

<Examples> Examples of the present invention will be described below.

In FIG. 2 showing one embodiment, a knock sensor (vibration sensor) 1 attached to a cylinder block of an internal combustion engine (not shown) has a built-in thick electric element and outputs a detection (voltage) signal having a waveform corresponding to engine vibration. Output.

The detection signal (analog signal) of the knock sensor 1 is A
It is A / D converted by the / D converter 2 and input to the comb filter 3.

The comb filter 3 is composed of a delay circuit 4 composed of a plurality of stages of unit delay elements, and an adder 5 for subtracting the output data of the delay circuit 4 from the data bypassing the delay circuit 4. To the comb filter 3, a circuit in which a number of resonators 6a to 6e corresponding to the number of frequencies desired to be extracted from the detection signal of the knock sensor 1 are connected in parallel is vertically connected.

The resonators 6a to 6e are adapted to resonate with mutually different frequency components, and in the present embodiment, the resonance frequency is 7 kHz according to the frequency range 7 kHz to 9 kHz in which knocking vibration is said to be prominently expressed. , 8kH
z, 9 kHz, 10 kHz and 11 kHz.

In the comb filter 3, by subtracting the data delayed by the delay circuit 4 from the data bypassed by the delay circuit 4, the detection signal level is totally attenuated, and in particular, the frequency corresponding to the delay time is added. They are erased by the device 5 to obtain a so-called comb-shaped result as a frequency characteristic.

As a result, the resonators 6a to 6e can be prevented from continuing to resonate based on the signals canceled by the adder 5, and the intensities of the respective frequency components can be successively obtained. The comb-shaped filter 3 and the resonators 6a to 6e constitute the specific frequency component extracting means in this embodiment.

In the present embodiment, the specific frequency component is extracted from the signal from the knock sensor 1 by the configuration of the comb filter 3 and the resonators 6a to 6e as described above. It is also possible to provide the bandpass filters in correspondence with the number of required frequencies and A / D convert the output of each bandpal filter so that the microcomputer 7 can read the outputs.

The output of each of the resonators 6a to 6e, that is, the intensity signal of each frequency component is input to the microcomputer 7, and the microcomputer 7 is based on the detection signal from the crank angle sensor 8. The occurrence of knocking in an internal combustion engine (not shown) is detected based on a specific frequency component of the knock sensor 1 input through the resonators 6a to 6e in a predetermined frequency analysis section to be detected.

The content of such knocking occurrence detection will be described below with reference to the flowchart of FIG. In this example,
As shown in the flow chart of FIG. 3, the microcomputer 7 is provided with software as functions of intensity sampling means, maximum level calculation means, average level calculation means, knocking determination means and background level calculation means. The frequency contribution rate storage means is composed of a nonvolatile memory built in the microcomputer 7.

The knocking occurrence detection shown in the flowchart of FIG.
This is performed in a predetermined frequency analysis section, and the predetermined frequency analysis section is, for example, a section in which combustion vibration of each cylinder can be sampled while avoiding ignition noise. For example, in a 6-cylinder engine, ATDC 10 ° to ATDC 60 °. Then, the predetermined frequency analysis section is detected based on the detection signal from the crank angle sensor 8.

When the frequency analysis section is entered, the frequency spectra as shown in FIG. 7 to be output from the resonators 6a to 6e at every predetermined time (predetermined period) are sequentially stored, so that as shown in FIG. 4 and FIG. In addition, the intensity (f ₀₀ , f 0, n) of each frequency component f _j (j = 0 to n) obtained at every predetermined time in the analysis section.
f ₁₀ , f ₂₀ , f _30, ... F _m0 ), (f ₀₁ , f ₁₁ , f ₂₁ , f
₃₁ ...... f _m1 ), (f ₀₂ , f ₁₂ , f ₂₂ , f ₃₂・ ・_・ f _m2 ) ...
.. (f _0n , f _1n , f _2n , f _2n , f _3n, ... F _mn ) are sampled (S1). This function corresponds to intensity sampling means.

Then, based on these sampled intensity data, the maximum intensity level KSi of all frequencies is calculated by the following equation (S2). This function corresponds to the maximum level calculation means.

KSi = f _m0 (max) × KF ₀ + f _m1 (max) × KF ₁ ＋ …… ＋ f
_mn (max) × KF _n Here, f _mj (max) is the maximum value among the sampling intensities (f _0j , f _1j , f _2j , f _3j ... f _mi ) of the frequency component f _j , and KF _i Is a frequency contribution rate indicating the degree of contribution of the frequency component f _i to knocking. Since the frequency contribution rate KF _i varies depending on the cylinder and the engine rotation speed, the accuracy can be improved by searching for the cylinder and the engine rotation speed from those stored in the three-dimensional map.

This maximum level KSLi is a value that serves as an index indicating the magnitude of knocking.

Next, based on the intensity data, the average intensity level SLi of all frequencies is calculated by the following equation (S
3). This function corresponds to the average level calculation means.

SLi = {KF ₀ / m Σf _m0 ＋ KF ₁ / mΣf _m1 ＋ …… KF _n / mΣ
f _mn } maximum level KSLi and average level thus obtained
The presence or absence of knocking is determined by comparing the difference with SLi with the background BGLi (S4). That is, BGLi + α <KSi
When −SLi, it is determined that knocking has occurred (S5), and BGLi ≧ KSLi−SLi + α (α is a threshold value)
If it is, it is determined that knocking has not occurred (S6). This function corresponds to knocking determination means.

The background level BGLi is obtained by performing a weighted average calculation of the difference between the maximum level KSi and the average level SLi compared with the BGLi by the following equation when it is determined that there is no knocking in the knocking determination ( S7). This function corresponds to the background calculation means.

According to such knocking discrimination, the difference between the maximum level KSi and the average level SLi is basically the sum of the differences between the maximum value and the average value of the vibration level, and when knocking occurs, the above-mentioned knocking vibration is locally detected. This difference increases due to the property of increased vibration levels.

On the other hand, the background level BGLi is a weighted average value of the sum of the differences between the maximum level KSi and the average level SLi when not knocking.

Therefore, when knocking occurs, the value of (KSi-SLi) is sufficiently larger than that of BGLi, and accurate knocking determination can be performed even for a signal with a low level.

In addition, since a plurality of specific frequencies having a high contribution rate to knocking are selected and knocking determination is performed based on the sum of intensities, knocking detection accuracy is improved, and in particular, in the present embodiment, weighting is performed using a frequency contribution rate for each frequency component. The knocking detection accuracy is further improved because the sum is obtained by performing the calculation.

Next, an example will be described in which knocking vibration characteristics are more accurately captured to make the distinction from mechanical vibration more clear and knocking determination is performed.

In this method, when the average level SLi is calculated, the total sum of intensity average values for each frequency obtained in the same manner as in the above-described embodiment is corrected with a value that is an intensity change correction coefficient KLP. Also, when calculating BGLi, the weighting x is set as a function of the intensity change correction coefficient KLP. The routine of this embodiment for performing knocking detection using the intensity correction coefficient KLP will be described with reference to the flowchart of FIG.

First, the intensity for each frequency component is sampled at predetermined time intervals in the same manner as S1 in FIG. 3 (S11), and then the intensity of each specific frequency component (f ₀₀ , f ₁₀ ... .. (f _n0 ) are stored as initial values (S12).

Then, it is assumed that the initial values stored for each frequency component do not change, respectively, and that the intensity of a constant level continues in the frequency analysis section, and the time axis change of the integrated value of the intensity at this time is compared with the sampling period and the initial value. The value is set on the basis of the value and it is set as a reference change characteristic (S13).

Next, based on the temporal transition of the intensity actually obtained for each frequency component (see FIG. 7), the intensity is integrated for each sampling period on the time axis, and the intensity within the frequency analysis section is calculated. The change characteristic is detected (S14). This function corresponds to intensity change detection means.

Then, as described above, the reference change characteristic for each frequency component obtained assuming that the intensity is unchanged is compared with the actually detected change characteristic of the intensity integral value for each frequency component.
(S15).

Here, the reference change characteristic is linearly increased because it is premised that there is no change in intensity. On the other hand, when the change characteristic of the intensity integral value obtained based on the actual detection signal does not match. Is estimated to be unstable at a constant intensity because the frequency component of the noise includes knocking vibration. Therefore, it is determined whether or not the intensity integral value matches the reference variation characteristic for each predetermined time for each frequency component, and the total number C in the case where they do not match within the detection section is measured by the counter.

Then, the intensity change correction coefficient KLP is set according to the total number C. Specifically, the larger the total number C, the greater the proportion of knocking vibrations included, and therefore the direction in which knocking determination is likely to occur, that is, the average level SLi is reduced to a small value to maximize the difference from the maximum level KSi. Yes (S16).

The intensity change correction coefficient calculation means is composed of S15 and the function of S16.

Then, after the maximum intensity level KSi is calculated in the same manner as S2 in FIG. 3 (S17), the average level SLi is calculated by the following equation using the intensity change correction coefficient KLP set as described above (S18). ).

SLi = {KF ₀ / m Σf _m0 ＋ KF ₁ / m Σf _m1 ＋ …… KF _n / mΣf
_mn } ... KLP Next, the difference between the maximum level KSi and the average level SLi is compared with the background level BGLi to determine the presence or absence of knocking (S19). (S20), BGLi + α ≧ KSi-SLi
If it is, it is determined that there is no knocking (S21).

When it is determined that there is no knocking, the background level BGLi is calculated and updated, but before that, the weighting coefficient x used for the calculation is set as a function of the intensity correction coefficient KLP. Specifically, the larger KLP is, the smaller the proportion of knocking vibrations is.
Weighting coefficient x to calculate BGLi by increasing the weight of
Is set as a small value (S22).

Using x set in this way, S7 in FIG.
BGLi is calculated using the equation used in (S23).

As described above, by using the intensity correction coefficient KLP, it is possible to detect the intensity change characteristic of knocking and perform knocking detection with higher accuracy.

In this embodiment, the intensity change correction coefficient KLP is set as a value for correcting the average level, but it may be set as a value for correcting the maximum level.

<Effects of the Invention> As described above, according to the present invention, the intensities of a plurality of specific frequency components from the detection signal from the vibration sensor are sampled at predetermined intervals within a predetermined interval, and the maximum level and the average value are averaged. Since the knocking detection is performed based on the difference from the average level, which is a value, it is possible to perform highly accurate detection matching the knocking characteristic. Further, by performing the correction calculation using the frequency contribution rate and the intensity change correction coefficient, the effect that the knocking accuracy can be further enhanced is obtained.

[Brief description of drawings]

1 is a block diagram showing the configuration of the present invention, FIG. 2 is a system block diagram showing an embodiment of the present invention, FIG. 3 is a flowchart showing the contents of knocking detection control in the same embodiment, FIG. FIG. 5 is a time chart showing the state of intensity sampling for each specific frequency component in the same embodiment, FIG. 6 is a flowchart showing the contents of knocking detection control in another embodiment, and FIG. 7 is the same embodiment. FIG. 8 is a diagram showing an example of a frequency spectrum in FIG. 8, and FIG. 8 is a block diagram showing an example of a conventional knocking detection device. 1 ... Knock sensor (vibration sensor), 2 ... A / D converter, 3 ... Comb filter, 4 ... Delay circuit, 5 ... Adder, 6a-6e ... Resonator, 7 ... Micro Computer, 8 ... Crank angle sensor

Claims (4)

[Claims] 1. A vibration sensor attached to an engine body for detecting engine vibration, a specific frequency component extracting means for extracting a plurality of specific frequency components from a detection signal of the vibration sensor, and an extraction by the specific frequency component extracting means. Intensity sampling means for sampling the intensity of each of the plurality of specified frequency components for each predetermined period within a predetermined interval, and the sum of the maximum values of the intensities sampled in the predetermined interval of each specific frequency component as the maximum level. A maximum level calculating means for calculating, an average level calculating means for calculating a sum of average values of intensities sampled in a predetermined section of each specific frequency component as an average level, and a maximum calculated by the maximum level calculating means The difference between the level and the average level calculated by the average level calculating means is calculated as the background. A knocking determination means for determining presence or absence of knocking by comparing with a bell, and a difference between the maximum level calculated by the maximum level calculation means and the average level calculated by the average level calculation means, the knocking determination means knocks A knocking detection device for an internal combustion engine, comprising: a background level calculating means for calculating a weighted average value obtained by performing a weighted average as a background level when it is determined to be absent. 2. A frequency contribution rate storage means for storing a frequency contribution rate that contributes to knocking of the specific frequency component for each specific frequency component, wherein the maximum level calculation means and the average level calculation means are the maximum values of the specific frequency components. 2. The knocking detection device for an internal combustion engine according to claim 1, wherein the sum of values obtained by correcting the average value and the frequency contribution rate of the corresponding frequency component is calculated as the maximum level and the average level. 3. An intensity change correction means for detecting an intensity change for each specific frequency component, and an intensity change correction by comparing the intensity change of the specific frequency component detected by the intensity change detection means with a predetermined reference change characteristic. An intensity change correction coefficient calculation means for calculating a coefficient, and the average level calculated by the average level calculation means or the maximum level calculated by the maximum level calculation means is obtained by correcting the intensity change correction coefficient. Item 3. A knocking detection device for an internal combustion engine according to item 1 or 2. 4. The internal combustion engine according to claim 3, wherein the weighting coefficient used in the background level calculation means for calculating the background level is set as a function of the intensity correction coefficient calculated by the intensity change correction coefficient calculation means. Knocking detection device.