JP5927740B2 - Knock detection device - Google Patents

Description

  The present invention relates to a knock detection device, and more particularly to a knock control device for an engine, and more specifically, from a signal detected by a knock sensor installed on the cylinder block surface of the engine, a signal other than a knock and a signal based on the knock The present invention relates to a knock detection device that accurately detects the occurrence of a knock by discriminating a noise signal originating from a mechanical signal.

Generally, when knocking occurs in an engine, vibrations having a plurality of resonance frequencies specific to the engine are generated.
In order to detect this knock, a knock sensor is installed in the cylinder block of the engine.
At this time, in order to perform highly accurate knock detection, the pass frequency of the bandpass filter and the window extraction time zone regarding the knock sensor are important.

JP 2005-188297 A JP 2009-209683 A JP 2011-179320 A JP 2011-179321 A JP 2011-179322 A JP 2011-179323 A

By the way, in the conventional knock detection device, in general knock control, a vibration is detected by a knock sensor installed in a cylinder block of an engine, and a band-pass filter for extracting a knock-specific frequency band from the knock sensor signal, and Applying time-domain window processing to extract only the signal in the knock occurrence time zone, and using the maximum value or integral value of the obtained signal waveform as an index of knock intensity, knock occurrence is recognized when it exceeds a certain threshold Therefore, knocking is suppressed by retarding the ignition timing.
At this time, since the vibration input to the knock sensor includes various mechanical vibrations, if only the component caused by the knock cannot be accurately detected from the signal, the knock cannot be sufficiently suppressed, and the engine Reduced product quality due to reduced durability or a knocking sound to the driver, or decreased output and fuel consumption performance due to erroneous detection of mechanical vibration and retarded ignition timing, or excessive increase in exhaust temperature There is a problem of inviting.

As a knock detection method for improving such a problem, a “phenomenon” in which the knock frequency shifts to the low frequency side with the lapse of time is used, and if the slope of the change exceeds a predetermined value, it is determined that the knock is detected. The method is disclosed in Patent Document 1 and Patent Document 2 described above.
In addition, it is because it is known that the knock frequency causes a frequency shift to the low frequency side with the passage of time if the above-mentioned “phenomenon” is added. On the other hand, the noise frequency due to mechanical vibration is low frequency. The frequency shift to the side is not caused.
However, there are problems shown in the following (1) and (2).
(1) Although the knock frequency always shifts to the low frequency side, the frequency shift may not occur for a certain period immediately after the occurrence of the knock.
In this case, there is a disadvantage that it is difficult to calculate the slope of the frequency change.
(2) Including the case of (1) above, there is an inconvenience that it becomes difficult to set a predetermined value (threshold value) for determining the occurrence of knocking when the slope of the knock frequency change becomes small.
And the method of dealing with said (1) and (2) is not disclosed by the above-mentioned patent document 1 and patent document 2.
As a result, Patent Document 1 and Patent Document 2 have insufficient knock detection strategies and have room for improvement.

  According to the present invention, a combustion cylinder (knock occurrence) is determined by simultaneously measuring three types of waveforms of a knock sensor waveform, a cam angle sensor waveform, and a crank angle sensor waveform, and the knock sensor waveform is cranked. The accuracy of knock occurrence is improved by performing frequency analysis in an angle-based time series and by not using the index of knock occurrence as the slope of the change in peak frequency but using the shift amount of the peak frequency and the ratio of the peak frequency. The purpose is to let you.

Therefore, in order to eliminate the above-described disadvantages, the present invention is a knock sensor that is attached to an engine and detects vibration, a cam angle sensor that is attached to the engine and detects a cam angle, and a crank angle that is attached to the engine and detects a crank angle. In a knock detection device comprising: a crank angle sensor, and a processing device that measures and processes a detection signal output from the knock sensor, the cam angle sensor, and the crank angle sensor as a waveform signal and determines whether knock has occurred or not The processing device comprises: a waveform measuring means for simultaneously measuring a knock sensor waveform from the knock sensor, a cam angle sensor waveform from the cam angle sensor, and a crank angle sensor waveform from the crank angle sensor; and A frequency analysis means for calculating a time-series frequency characteristic for the knock sensor waveform; Based on the frequency characteristics of the time series of the knock sensor waveform calculated knock frequency of several analysis means is shifted to a lower frequency with time, and the maximum value of the difference between the maximum value and the minimum value of the peak frequency or the peak frequency, And a knock occurrence determination means for determining that a knock has occurred when the ratio exceeds a predetermined threshold value.

According to the present invention, the combustion cylinder (knock occurrence) can be determined by simultaneously measuring three types of waveforms of the knock sensor waveform, the cam angle sensor waveform, and the crank angle sensor waveform. .
Further, it is possible to perform frequency analysis of the knock sensor waveform in a time series based on the crank angle.
Furthermore, the accuracy of determination of knock occurrence can be improved by not using the index of occurrence of knock as the slope of the change in peak frequency but using the peak frequency shift amount and the peak frequency ratio.

FIG. 1 is a flowchart for controlling the knock detection apparatus according to the first embodiment of the present invention. Example 1 FIG. 2 is a block diagram of the knock detection device. Example 1 FIG. 3 is a diagram showing a sensor attached to the engine and a detected signal. Example 1 FIG. 4 is a diagram illustrating setting of a time-series window based on a crank angle. Example 1 FIG. 5 is a diagram showing an example of time-series frequency analysis results with n peak frequencies. Example 1 FIG. 6 is a diagram showing a time-series frequency analysis result (no knock occurrence). Example 1 FIG. 7 is a diagram showing time-series frequency analysis results (knock occurrence). Example 1 FIG. 8 is a diagram showing the value of the shift amount f (i) s of each peak frequency in the first embodiment. Example 1 FIG. 9 is a diagram illustrating an example in which a knock frequency shift occurs immediately after the occurrence of a knock. Example 1 FIG. 10 is a diagram illustrating an example in which a knock frequency shift occurs after a certain time has elapsed. Example 1 FIG. 11 is an explanatory diagram of the tendency of the spectrum peak due to knocking according to the second embodiment of the present invention. (Example 2) FIG. 12 is an explanatory diagram of spectral peaks caused by knocking and noise. (Example 2) FIG. 13 is a block diagram of a knock detection apparatus showing a third embodiment of the present invention. (Example 3)

  Embodiments of the present invention will be described below in detail with reference to the drawings.

1 to 10 show an embodiment of the present invention.
In FIG. 2, reference numeral 1 denotes a knock detection device.
As shown in FIG. 2, the knock detection device 1 is attached to the engine 2 to detect a vibration, a knock sensor 3 that is attached to the engine 2 to detect a cam angle, and is attached to the engine 2. A crank angle sensor 5 for detecting a crank angle, and a process for measuring the detection signals output from the knock sensor 3, the cam angle sensor 4, and the crank angle sensor 5 as waveform signals and processing them to determine whether knocks have occurred or not. And a device 6.
The knock sensor 3 detects vibration on the surface of the cylinder block 2 a of the engine 2.
The cam angle sensor 4 detects a rotation angle of a cam shaft (not shown).
Further, the crank angle sensor 5 detects a rotation angle of a crankshaft (not shown).

At this time, the processing device 6 includes a waveform measuring means 7, a frequency analyzing means 8, and a knock occurrence determining means 9 as shown in FIG.
More specifically, the waveform measuring means 7 uses a data recorder (not shown) connectable to a personal computer (also referred to as “PC”) (not shown), the knock sensor waveform from the knock sensor 3 and the The cam angle sensor waveform from the cam angle sensor 4 and the crank angle sensor waveform from the crank angle sensor 5 are simultaneously measured.
The frequency analysis means 8 calculates time-series frequency characteristics for the knock sensor waveform of the waveform measurement means 7.
That is, the frequency analysis means 8 analyzes the knock sensor waveform obtained at the timing of combustion in time series.
Further, the knock occurrence determination means 9 determines whether the knock frequency calculated by the frequency analysis means 8 shifts the peak frequency or the peak frequency based on the time-series frequency characteristics of the knock sensor waveform in which the knock frequency shifts to the low frequency side as time elapses. Using the frequency ratio as an index, it is determined that knocking has occurred when the index exceeds a predetermined threshold.
That is, in the knock occurrence determination means 9, from the time-series frequency analysis result obtained by the frequency analysis means 8, for example, the shift amount of each peak frequency is used as an index, and a spectrum peak whose index exceeds a predetermined threshold is detected. If it exists, it is determined as a knock.

In addition, as shown in FIG. 2, the processing device 6 has three types of sensors, that is, the knock sensor 3, the cam angle sensor 4, and the crank angle sensor 5 attached to the engine 2. As shown in the figure, a signal is detected by each sensor.
At this time, when simultaneously measuring the waveforms detected by the three types of sensors by the waveform measuring means 7 of the processing device 6, it is necessary to determine the measurement conditions (engine speed, cylinder) of the simultaneously measured waveforms from the measured waveforms. There is.
The engine speed can be determined by the crank angle sensor 5, and the combustion of the cylinder, that is, the occurrence of knocking is determined by the waveform of the crank angle sensor 5 and the waveform of the cam angle sensor 4. Cylinder discrimination is possible.

In addition, this invention implement | achieves high knock detectability by solving the problems (1) and (2) described in the subject mentioned above.
If it adds, the example which frequency-analyzed the knock waveform detected with the knock sensor in FIG.9 and FIG.10 for every crank angle area will be shown.
FIG. 9 is an example in which the knock frequency shifts to the low frequency side immediately after the occurrence of the knock,
FIG. 10 is an example in which the knock frequency shifts to the low frequency side after a certain time has elapsed after the occurrence of the knock.
At this time, the solid line in the figure indicates the knock frequency immediately after the occurrence of the knock, and the dotted line indicates the state of the knock frequency shift.
The calculation of the slope of the frequency change of the knock shown in Patent Document 1 described above is relatively easy in the case shown in FIG. 9, but in the case shown in FIG. 10, there is a region where the frequency shift occurs and a region where it does not occur. It becomes difficult.
Even if the slope of FIG. 10 is calculated by the least square method or the like as shown in Patent Document 2 described above, the slope value is smaller than in the case of FIG. It becomes difficult to set a predetermined threshold value.
Therefore, in order to deal with these problems, the present invention proposes a method in which an index for determining the occurrence of knocking is used as a knock frequency shift amount.
According to this method, the value of the knock frequency shift amount (f _max −f _min ) is clear also in FIG. 10, and the setting of the predetermined threshold value is facilitated.
Further, the calculation of the amount of shift of the knock frequency has an effect that the calculation processing is less than the calculation of the slope of the change of the knock frequency.

Hereinafter, the frequency analysis unit 8 and the knock occurrence determination unit 9 of the processing device 6 will be described in more detail.
The frequency analysis means 8 performs a frequency analysis in a time series based on the crank angle based on the crank angle sensor waveform for the knock sensor waveform obtained at the timing of combustion.
In the time-series frequency analysis by the frequency analysis means 8, the time axis of the time-series frequency analysis is a crank angle often used in knock control. For example, the time axis may be an elapsed time from the ignition timing. good.
As shown in FIG. 4, windows are set in time series with respect to the knock sensor waveform (in this example, eight windows (a) to (h) are set every 10 degrees of crank angle, and the window width is 30 degrees of crank angle). ), Frequency characteristics are obtained for each window using FFT or DFT.
At this time, the window width and the interval between adjacent windows are not limited, but it is necessary to consider the range from the occurrence of knocking to the sufficient attenuation of knocking.
In the setting of the window shown in FIG. 4, for example, a crank angle-based time-series frequency analysis result as shown in FIGS. 9 and 10 is obtained.
Further, the knock occurrence determination means 9 uses the shift amount of each peak frequency as an index from the time-series frequency analysis result obtained by the frequency analysis means 8, and there is a spectrum peak whose index exceeds a predetermined threshold. If it is knocked.
The knock occurrence determination means 9 calculates the shift amount of the peak frequency based on the difference between the maximum value and the minimum value of the peak frequency, while calculating n indexes for the N peak frequencies, and the n indexes. When at least one of them exceeds a predetermined threshold value, it is determined that knocking has occurred.
That is, as shown in FIG. 5, it is assumed that N peak frequencies f (i) (i = 1 to n) are detected.
For the peak frequency f (i), the maximum frequency f (i) _{max that} is the maximum value and the minimum frequency f (i) _min that is the minimum value are obtained from all the windows (five windows in FIG. 5). The peak frequency shift amount f (i) _s (f (i) _s = f (i) _max −f (i) _min ) is calculated.
However, since the spectrum peak due to knock shifts to the low frequency side with time, when the spectrum peak shifts to the high frequency side, that is, the minimum frequency f (i) _min is earlier than the maximum frequency f (i) _max . When it is obtained from the window, the peak frequency shift amount f (i) _{s is set} to 0 (f (i) _s = 0).
For the N peak frequencies f (i) (i = 1 to n), the peak frequency shift amount f (i) _s (i = 1 to n) is calculated, and the index f (i) _s (i = 1 to n) is calculated. ) Is judged as knocking when a predetermined threshold value is exceeded.

Here, using two knock sensor waveforms whose knock occurrence and knock occurrence are known in advance, it is confirmed whether the occurrence of knock can be determined by the proposed measure.
FIG. 6 shows the crank angle base by the frequency analysis means 8 for the knock sensor waveform when the ignition timing is sufficiently retarded and knock does not occur in a certain engine (rotation speed 2400 rpm, combustion of the third cylinder). This is the result of frequency analysis of time series.
From FIG. 6, two peak frequencies f (1) and f (2) are detected.
In these peak frequencies f (1) and f (2), since the spectrum peak is slightly shifted to the high frequency side as time passes, the minimum frequency f (i) _min is the previous window from the maximum frequency f (i) _max. The peak frequency shift amounts f (1) _s and f (2) _s are both 0 (f (1) _s = 0, f (2) _s = 0).
FIG. 7 shows a crank angle-based time-series frequency analysis performed by the frequency analysis means 8 on the knock sensor waveform when the ignition timing is sufficiently advanced in the same engine (same condition) and knock occurs. It is a result.
From FIG. 7, four peak frequencies f (1) to f (4) are detected.
At this time, the peak frequencies f (1) and f (2) are the same as those detected in FIG. 6, and are peaks due to noise that is constantly generated.
Therefore, as in FIG. 6, the peak frequency shift amounts f (1) s and f (2) s are both 0 (f (1) _s = 0, f (2) _s = 0).
On the other hand, the peak frequencies f (3) and f (4) are shifted to the low frequency side as time passes, and it can be seen that they are peaks due to knocking.
At this time, for the peak frequencies f (3) and f (4), the values of the maximum frequency f (i) _max and the minimum frequency f (i) _min are respectively
f (3) _max = 7.52 (kHz), f (3) _min = 6.84 (kHz)
f (4) _max = 8.54 (kHz), f (4) _min = 7.52 (kHz)
It becomes.
Accordingly, the value of the peak frequency shift amount f (i) _s is respectively
f (3) _s = 0.68 (kHz)
f (4) _s = 1.02 (kHz)
It becomes.
The value of the shift amount f (i) _s of the peak frequency, which is an index for determining the occurrence of knocking as described above, is organized and disclosed in FIG.
From FIG. 8, if the predetermined threshold value for determining the occurrence of knocking is set larger than 0 (kHz) and smaller than 1.02 (kHz), the occurrence of knocking can be determined by the proposed measure in this embodiment. .
However, with regard to the predetermined threshold, there is a possibility that the peak frequency shift amount f (i) _s due to noise may be slightly larger than 0 due to an error in calculating the peak frequency, and the peak frequency shift amount due to knocking. f (i) It is necessary to set in consideration of the possibility that _s becomes smaller than that of the embodiment.

  Next, the operation will be described along the control flowchart of the knock detection device 1 of FIG.

When the control program of the knock detection device 1 starts, that is, when the processing starts (101), the processing shifts from time series frequency analysis to processing (102) for selecting the peak frequency f (i) (i = 1 to n). .
In this processing (102), the frequency analysis means 8 selects the peak frequency f (i) (i = 1 to n) from the time-series frequency analysis for the knock sensor waveform of the waveform measurement means 7. is there.
Then, after the process (102) of selecting the peak frequency f (i) (i = 1 to n) from the time series frequency analysis, the process proceeds to the process (103) of calculating the peak frequency of one window.
This process (103) is to calculate the peak frequency of one window of the peak frequency f (i) selected by the frequency analysis means 8, that is, to calculate the frequency characteristic.
In addition, after the process (103) of calculating the peak frequency of one window, the process proceeds to determination (104) of whether or not the calculation of the peak frequency of all windows is completed.
If this determination (104) is NO, the processing returns to the above-described processing (103) for calculating the peak frequency of one window.
In the determination (104) of whether or not the calculation of the peak frequency of all the windows is completed, if this determination (104) is YES, the maximum frequency f (i) _max that is the maximum value of the peak frequency f (i). And the process (105) for selecting the minimum frequency f (i) _min which is the minimum value.
After this processing (105), the peak frequency shift amount f (i) _s is _expressed by the equation f (i) _s = f (i) _max -f (i) _min
The process proceeds to the processing (106) calculated by the following.
In this process (106), the knock occurrence determination means 9 calculates the peak frequency shift amount f (i) _s from the equation f (i) _s = f (i) _max -f (i) _min.
Is used as an index.
Further, the peak frequency shift amount f (i) _s is _expressed by the following equation: f (i) _s = f (i) _max −f (i) _min
After the processing calculated by (106), the process proceeds to determination (107) as to whether or not calculation of n indices for N peak frequencies has been completed.
This determination (107) is based on an index f (i) _s (i = 1 to n) which is a shift amount of n peak frequencies with respect to N peak frequencies f (i) (i = 1 to n). It is determined whether or not the calculation of has been completed.
If the determination (107) is NO, the processing returns to the processing (102) for selecting the peak frequency f (i) (i = 1 to n) from the above time-series frequency analysis.
Further, in the determination (107) of whether or not the calculation of n indices for the N peak frequencies is completed, if the determination (107) is YES, the index f (i) _s (i = 1 to n). ), The process proceeds to the determination (108) of whether or not a threshold value that is a predetermined value is exceeded.
This determination (108) is for determining whether at least one of the calculated indices f (i) _s (i = 1 to n) exceeds a predetermined threshold in the knock occurrence determination means 9. is there.
If the determination (108) is YES, the process proceeds to a process (109) for determining that the knock occurrence determination means 9 is “knock”, that is, a knock has occurred.
Further, in the determination (108) of whether or not any one of the indices f (i) _s (i = 1 to n) exceeds a predetermined threshold value, when the determination (108) is NO, The knock occurrence determination means 9 shifts to a process (110) for determining that “no knock”, that is, no knock has occurred.

According to the first embodiment, the processing device 6 of the knock detection device 1 includes the knock sensor waveform from the knock sensor 3, the cam angle sensor waveform from the cam angle sensor 4, and the crank angle sensor 5. A waveform measuring means 7 for simultaneously measuring the crank angle sensor waveform, a frequency analyzing means 8 for calculating a time-series frequency characteristic for the knock sensor waveform of the waveform measuring means 7, and a knock frequency calculated by the frequency analyzing means 8 Based on the time-series frequency characteristics of the knock sensor waveform that shifts to the low frequency side as time passes, the amount of peak frequency shift is used as an index, and knocking occurs when the index exceeds a predetermined threshold. And a knock occurrence determining means 9 for determining.
Therefore, the combustion cylinder (knock occurrence) can be determined by simultaneously measuring three types of waveforms of the knock sensor waveform, the cam angle sensor waveform, and the crank angle sensor waveform.
Further, it is possible to perform frequency analysis of the knock sensor waveform in a time series based on the crank angle.
Furthermore, the accuracy of determination of knock occurrence can be improved by not using the index of occurrence of knock as the slope of the change in peak frequency but using the peak frequency shift amount and the peak frequency ratio.

The knock occurrence determination means 9 calculates the shift amount of the peak frequency based on the difference between the maximum value and the minimum value of the peak frequency, while calculating n indexes for the N peak frequencies. When at least one of the indices exceeds a predetermined threshold, it is determined that knocking has occurred.
Therefore, n indexes for N peak frequencies are calculated from the shift amount of the peak frequency, and when at least one of the n indexes exceeds a predetermined threshold, it is determined that knocking has occurred. The occurrence can be accurately determined, which can contribute to the improvement of the reliability of the determination.

11 and 12 show a second embodiment of the present invention.
In the second embodiment, portions that perform the same functions as those of the first embodiment will be described with the same reference numerals.

  In the first embodiment, the shift amount of the peak frequency is used as an index. However, the feature of the second embodiment is that the ratio of peak frequencies is used as an index.

従って、ピーク周波数の比ｆ（ｉ）_Ｒを指標として用いると、ノックに起因するピーク周波数ｆ（１）、ｆ（２）、…とｆ（ｎｏｉｓｅ）との区別が可能となり、ｆ（ｎｏｉｓｅ）をノイズと判定することが可能である。 That is, the knock occurrence determination means calculates the peak frequency ratio f (i) _R for each peak frequency, and uses f (i) _R as an index for determining knock occurrence.
At this time, the peak frequency ratio f (i) _R is expressed by the equation f (i) _R = f (i) _max / f (i) _min.
Calculate by
In addition, since the process similar to the description described in the above-mentioned 1st Example is implemented except using the ratio of peak frequency as a parameter | index, description is abbreviate | omitted.
When this second embodiment is additionally described, as shown in FIG. 11, the slope of the change in peak frequency becomes smaller as the spectral peak caused by knocking has a lower frequency.
For this reason, in consideration of the case where only the lowest frequency peak frequency f (1) is detected when the conventional technique is used, the threshold value, which is a predetermined value for knock determination, is the lowest frequency peak frequency f (1). ) Slope S1 (1) needs to be set slightly smaller.
Here, as shown in FIG. 12, due to the variation of the peak frequency on the higher frequency side than the lowest frequency peak frequency f (1), it is caused by the noise having the same slope as the lowest frequency peak frequency f (1). Assume that a peak frequency f (noise) appears.
In this case, even if the knocking non-occurrence state in which the lowest peak frequencies f (1) and f (2) do not appear, there is a problem that the knocking is erroneously determined by the peak frequency f (noise).
Thus, this problem can be avoided if the index for determining knock occurrence is the peak frequency ratio f (i) _R.
First, peak frequency ratios (f (1) _R , f (2) _R ,...) Resulting from knocking basically take the same value α.
On the other hand, the peak frequency ratio f (noise) _R (f (noise) _max / f (noise) _min ) with respect to f (noise) having a slope equivalent to the lowest frequency peak frequency f (1) is expressed by the following equation: The value is clearly smaller than α.

Therefore, if the peak frequency ratio f (i) _R is used as an index, it becomes possible to distinguish the peak frequencies f (1), f (2),..., And f (noise) caused by knocking, and f (noise) Can be determined as noise.

Here, additional explanation will be given with respect to the above-mentioned point that “the ratios of peak frequencies (f (1) _R , f (2) _R ,...) Resulting from knock basically take the same value α”.
The peak frequency resulting from knocking is basically proportional to the sound velocity of the gas.
After knocking, the temperature of the gas decreases as the piston descends, and the sound velocity of the gas decreases as the temperature of the gas decreases.
Therefore, the peak frequency resulting from knocking changes to the low frequency side with time.
Here, since the peak frequency due to knock is proportional to the sound velocity of the gas, the peak frequencies f (1) _max and f (2) _max in FIG.
f (1) _max = Av
f (2) _max = Bv
(A, B: proportional constant, v: sound velocity of gas in (b))
far.
If the sound velocity v of the gas changes to v / α (1 <α) when time has elapsed since (b) and the state becomes (f), the peak frequencies f (1) _min and f (2) _min is f (1) _min = Av / α
f (2) _min = Bv / α
It becomes.
Therefore, the ratio of peak frequencies is f (1) _max / f (1) _min = α
f (2) _max / f (2) _min = α
Thus, it can be seen that the ratios of the respective peak frequencies resulting from knocking have the same value.

In other words, the processing device 6 of the knock detection device 1 receives the knock sensor waveform from the knock sensor 3, the cam angle sensor waveform from the cam angle sensor 4, and the crank angle sensor waveform from the crank angle sensor 5. The waveform measuring means 7 for simultaneously measuring, the frequency analyzing means 8 for calculating the time-series frequency characteristics for the knock sensor waveform of the waveform measuring means 7, and the knock frequency calculated by the frequency analyzing means 8 decreases with time. Based on the time-series frequency characteristics of the knock sensor waveform shifted to the frequency side, the ratio of peak frequencies is used as an index, and knock occurrence determination means for determining that knock has occurred when the index exceeds a predetermined threshold 9 and.
Therefore, the combustion cylinder (knock occurrence) can be determined by simultaneously measuring three types of waveforms of the knock sensor waveform, the cam angle sensor waveform, and the crank angle sensor waveform.
Further, it is possible to perform frequency analysis of the knock sensor waveform in a time series based on the crank angle.
Furthermore, the accuracy of determination of knock occurrence can be improved by not using the index of occurrence of knock as the slope of the change in peak frequency but using the peak frequency shift amount and the peak frequency ratio.

  FIG. 13 shows a third embodiment of the present invention.

  A feature of the third embodiment is that instead of the knock sensor, an in-cylinder pressure sensor 11 is installed in each cylinder to measure the in-cylinder pressure waveform.

That is, as shown in FIG. 13, the knock detection device 1 includes a cam angle sensor 4 that is attached to the engine 2 and detects a cam angle, a crank angle sensor 5 that is attached to the engine 2 and detects a crank angle, And a processing device 6 that measures and processes detection signals output from the knock sensor 3, cam angle sensor 4, and crank angle sensor 5 as waveform signals to determine whether or not knocking has occurred.
The cam angle sensor 4 detects the rotation angle of a cam shaft (not shown).
The crank angle sensor 5 detects the rotation angle of a crankshaft (not shown).
At this time, the processing device 6 includes a waveform measuring means 7, a frequency analyzing means 8, and a knock occurrence determining means 9 as shown in FIG.
As shown in FIG. 13, the knock detection device 1 is provided with, for example, four first to fourth in-cylinder pressure sensors 11 a to 11 d for each cylinder of the engine 2. The in-cylinder pressure waveform of each cylinder is measured by the in-cylinder pressure sensors 11a to 11d.
At this time, the frequency analysis means 8 and the knock occurrence determination means 9 may be processed in the same manner as the knock sensor.

In other words, instead of the knock sensor, the cylinder pressure sensor 11 is installed in each cylinder so that the cylinder pressure waveform is measured, so that it is possible to determine the cylinder that is burning (knock occurrence). is there.
In addition, the accuracy of determination of knock occurrence can be improved by not using the index of occurrence of knock as the slope of the change in peak frequency but using the shift amount of peak frequency and the ratio of peak frequencies.

DESCRIPTION OF SYMBOLS 1 Knock detection apparatus 2 Engine 3 Knock sensor 4 Cam angle sensor 5 Crank angle sensor 6 Processing apparatus 7 Waveform measurement means 8 Frequency analysis means 9 Knock generation determination means

Claims (2)

A knock sensor that is attached to the engine and detects vibration, a cam angle sensor that is attached to the engine and detects a cam angle, a crank angle sensor that is attached to the engine and detects a crank angle, and the knock sensor and the cam angle sensor And a processing device that measures and processes a detection signal output from the crank angle sensor as a waveform signal and determines whether knock has occurred or not, wherein the processing device includes a knock sensor waveform from the knock sensor. And waveform measuring means for simultaneously measuring the cam angle sensor waveform from the cam angle sensor and the crank angle sensor waveform from the crank angle sensor, and calculating time-series frequency characteristics for the knock sensor waveform of the waveform measuring means. The frequency analysis means and the knock frequency calculated by the frequency analysis means Based on the frequency characteristics of the time series of the knock sensor waveform to shift to a lower frequency, the difference between the maximum value and the minimum value of the peak frequency or an index ratio of the maximum value and the minimum value of the peak frequency, the index A knock detection device comprising: knock generation determination means for determining that a knock has occurred when the value exceeds a predetermined threshold.   The knock occurrence determination means of the processing device calculates the shift amount of the peak frequency based on the difference between the maximum value and the minimum value of the peak frequency, while calculating n indexes for the N peak frequencies, The knock detection device according to claim 1, wherein knocking is determined to occur when at least one of the indices exceeds a predetermined threshold.

Images