JP2008157087A - Knock detection device - Google Patents

Abstract

Engine knocking is detected with higher accuracy.
A knocking detection device includes: an in-cylinder pressure detecting unit that outputs a signal corresponding to an in-cylinder pressure of an engine; a removing unit that removes an in-cylinder pressure component in a knocking detection section from an output signal of the in-cylinder pressure sensor; Extraction means for extracting a signal in a predetermined frequency band including knocking frequency from the signal from which the in-cylinder pressure component has been removed, and determination means for determining occurrence of knocking from the extracted signal. Since the in-cylinder pressure component in the detection section is removed, knocking can be performed without being affected by the in-cylinder pressure component even when the detection section is shortened or time series frequency decomposition with a rough frequency characteristic is used for the extraction means. Can be detected.
[Selection] Figure 8

Description

  The present invention relates to an apparatus for detecting knocking of an internal combustion engine.

  Conventionally, an internal combustion engine (hereinafter referred to as an engine) is provided with a vibration sensor, and the occurrence of knocking is detected from the output of the vibration sensor.

According to the technique disclosed in Patent Document 1 below, the vibration sensor is provided in the cylinder of the engine. The wavelet transform is applied to the output of the vibration sensor to calculate the vibration frequency characteristics. Combustion failure such as knocking is detected by comparing the vibration frequency characteristic with a predetermined value.
JP 2002-221074 A

  The output of the vibration sensor may include noise when the intake / exhaust valve is closed (referred to as seating noise). In order to prevent such noise from being erroneously recognized as the occurrence of knocking, a method of setting a section for detecting knocking can be considered. In some cases, it may be desirable to set the detection interval to be short in order to separate the knocking signal from the seating noise well and to reduce the calculation load.

  On the other hand, a technique for extracting a signal including a knocking frequency from an in-cylinder pressure signal by a time-series frequency decomposition technique such as wavelet transform has been proposed. In order to reduce the calculation load, it may be desired to use time-series frequency decomposition with a rough frequency characteristic.

  However, when the detection interval is set short, or when time-series frequency decomposition with rough frequency characteristics is used, the influence of the in-cylinder pressure component may remain on the signal detected as knocking. The remaining in-cylinder pressure component may deteriorate the knocking detection accuracy.

  Accordingly, an object of the present invention is to provide an apparatus capable of detecting knocking with high accuracy even when time-series frequency decomposition with rough frequency characteristics is used and / or even when a detection interval is set to be short. It is.

  According to one aspect of the present invention, the knocking detection device includes an in-cylinder pressure detecting means for outputting a signal corresponding to the in-cylinder pressure that detects the in-cylinder pressure of the engine, and an output signal of the in-cylinder pressure sensor, from the knock detection interval. Removing means for removing the in-cylinder pressure component, extracting means for extracting a signal in a predetermined frequency band including the knocking frequency from the signal from which the in-cylinder pressure component has been removed, and determining the occurrence of knocking from the extracted signal Determining means.

  According to the present invention, since the in-cylinder pressure component in the detection section is removed, the influence of the in-cylinder pressure component can be achieved even when the detection section is shortened or when the time series frequency decomposition with a rough frequency characteristic is used for the extraction means. Knocking can be detected without receiving.

  According to one embodiment of the present invention, there is provided means for setting a knocking detection section based on information indicating engine rotation.

  According to the present invention, it is possible to appropriately set a section where knocking should be detected in accordance with the operating state of the engine.

  According to one embodiment of the present invention, there is provided means for detecting a peak of the in-cylinder pressure from an output signal of the in-cylinder pressure detecting means and setting a knocking detection section in a predetermined period from the peak point.

  According to the present invention, even when the engine has deteriorated over time or when there is a relative deviation between the peak timing of the in-cylinder pressure and the occurrence timing of knocking due to a variable compression ratio mechanism or the like, the knocking detection section is appropriately set. Can be set. In addition, the detection section can be set so as not to be affected by the seating noise.

  According to one embodiment of the present invention, the removing means removes the in-cylinder pressure component by a filter that cuts a low frequency component including the in-cylinder pressure component.

  According to the present invention, since the signal in the knocking frequency band is extracted after removing the frequency component including the in-cylinder pressure component, it is possible to avoid the in-cylinder pressure component from affecting the knocking detection.

  According to one embodiment of the present invention, the removing means calculates an approximate expression for approximating the in-cylinder pressure from the output signal of the in-cylinder pressure detecting means in the knocking detection section, and from the output signal of the in-cylinder pressure detecting means, By subtracting the in-cylinder pressure calculated by the approximate expression, the in-cylinder pressure component is removed. In one embodiment, the approximate expression is a linear expression. From the output signals of the in-cylinder pressure detecting means at the start point and the end point of the knocking detection section, the slope and intercept of the linear expression are obtained.

  According to the present invention, since the in-cylinder pressure component can be removed by a simple calculation, the calculation load can be reduced. Since the signal in the knocking frequency band is extracted after removing the frequency component including the in-cylinder pressure component, it is possible to avoid the in-cylinder pressure component from affecting the knocking detection.

  According to one embodiment of the present invention, the output signal of the in-cylinder pressure detecting means is a signal representing a differential value of the in-cylinder pressure.

  According to the present invention, knocking detection that is not affected by the in-cylinder pressure component can be performed using a signal (differential value) indicating the amount of change in the in-cylinder pressure.

  According to one embodiment of the present invention, if the time during which the signal extracted by the extraction unit is equal to or greater than a predetermined value continues for a predetermined time or more, it is determined that knocking has occurred. In another embodiment, the extracted signals are integrated over the knocking detection interval, and if the integrated value exceeds a predetermined value, it is determined that knocking has occurred.

  According to the present invention, knocking can be determined while reducing the influence of noise and the like.

  According to one embodiment of the present invention, the engine is an engine including a variable compression ratio mechanism that can change the compression ratio in the combustion chamber.

  Although the relative timing between the peak time of the in-cylinder pressure and the knocking occurrence time may be shifted by the variable compression ratio mechanism, knocking can be detected with good accuracy even in such a case.

1 is an overall configuration diagram of an internal combustion engine (hereinafter referred to as an engine) and a control device thereof according to an embodiment.

  An electronic control unit (hereinafter referred to as “ECU”) 1 is a computer including a central processing unit (CPU) and a memory. The memory can store a computer program for realizing various controls of the vehicle and data necessary for executing the program. The ECU 1 receives data sent from each part of the vehicle, performs calculation, and generates a control signal for controlling each part of the vehicle.

  The engine 2 has four cylinders in this embodiment, and only one of them is shown in the figure.

  The engine 2 is connected to an intake pipe 4 via an intake valve 3 and connected to an exhaust pipe 6 via an exhaust valve 5. A fuel injection valve 7 and a spark plug 9 are attached so as to face the combustion chamber 8. The combustion injection valve 7 injects fuel in accordance with a control signal from the ECU 1. The timing of injecting the fuel and the amount of fuel to be injected are controlled by the ECU 1 through the control signal. The spark plug 9 blows a spark in accordance with a control signal from the ECU 1. Due to the spark, the injected fuel / air mixture burns in the combustion chamber 8.

  Combustion increases the volume of the air-fuel mixture, thereby pushing the piston 10 downward. The reciprocating motion of the piston 10 is converted into the rotational motion of the crankshaft 11.

  The engine 2 is provided with a crank angle sensor 17. The crank angle sensor 17 outputs a CRK signal and a TDC signal to the ECU 1 as the crankshaft 11 rotates.

  The CRK signal is a pulse signal output at a predetermined crank angle. The ECU 1 calculates the rotational speed NE of the engine 2 according to the CRK signal. The TDC signal is a pulse signal output at a crank angle related to the TDC position of the piston 10.

  The in-cylinder pressure sensor 15 is a sensor that outputs a signal corresponding to the pressure in the combustion chamber 8 (in-cylinder pressure), and is buried in a portion of the spark plug 9 that contacts the engine cylinder. When the in-cylinder pressure sensor 15 is a sensor made of, for example, a piezoelectric element, the in-cylinder pressure sensor 15 outputs a signal dP indicating a change in pressure in the combustion chamber 8 (in-cylinder pressure), and the in-cylinder pressure change signal dP is For example, the in-cylinder pressure signal P indicating the in-cylinder pressure can be generated by integration in the ECU 1. Alternatively, the in-cylinder pressure sensor may be configured to include an integration operation (for example, incorporating an integration circuit).

  A throttle valve 18 is provided in the intake pipe 4 of the engine 2. The opening degree of the throttle valve 18 is controlled by a control signal from the ECU 1. A throttle valve opening sensor (θTH) 19 connected to the throttle valve 18 supplies an electrical signal corresponding to the opening of the throttle valve 18 to the ECU 1.

  The intake pipe pressure (PB) sensor 20 is provided on the downstream side of the throttle valve 18. The intake pipe pressure PB detected by the PB sensor 20 is sent to the ECU 1.

  An air flow meter (AFM) 21 is provided upstream of the throttle valve 18. The air flow meter 21 detects the amount of air passing through the throttle valve 18 and sends it to the ECU 1.

  The variable compression ratio mechanism 22 is a mechanism that can change the compression ratio in the combustion chamber 8 in accordance with a control signal from the ECU 1. The variable compression ratio (Cr) sensor 23 is a sensor for detecting the compression ratio Cr in the combustion chamber. The variable compression ratio mechanism 22 and the variable compression ratio sensor 23 can be realized by any appropriate technique. For example, a link type variable compression mechanism in which a piston and a crankshaft are connected by a plurality of links can be used. For example, a variable compression ratio sensor can be realized as a sensor that detects the compression ratio by detecting the rotation angle of the control shaft of the mechanism with a potentiometer (Japanese Patent Laid-Open No. 2006-226133). As an alternative, the variable compression ratio sensor can be realized as a potentiometer that detects the operating position of an actuator provided in the variable compression ratio mechanism. The potentiometer can detect a compression ratio (= in-cylinder pressure volume at intake bottom dead center / in-cylinder pressure volume at compression top dead center) from the operating position of the actuator (Japanese Patent Laid-Open No. 2003-193872).

  In this embodiment, the amount of intake air taken into the combustion chamber is controlled by the throttle valve, but may alternatively be controlled by a variable valve mechanism. The variable valve mechanism adjusts the intake air amount by changing the lift amount of the intake valve 3 in accordance with a control signal from the ECU 1.

  Here, how the in-cylinder pressure component may affect the detection of knocking will be described.

  It is known that the in-cylinder pressure change signal and the in-cylinder pressure signal obtained by integrating it include an in-cylinder pressure component, a knocking component, a seating noise component, and the like. The in-cylinder pressure component here is different from noise and indicates a frequency component constituting the in-cylinder pressure generated in the normal combustion state.

  FIG. 2 shows an example of the result of frequency analysis of the output signal dP of the in-cylinder pressure sensor by offline FFT. Reference numeral 25 indicates a frequency band of in-cylinder pressure, which is approximately 0 to 2 kHz. Reference numeral 26 denotes a knocking frequency band, which is about 6 to 8 kHz, depending on the bore diameter of the engine. Both frequency bands are very different. Therefore, the in-cylinder pressure component and the knocking component can be separated by frequency-decomposing the in-cylinder pressure change signal (or in-cylinder pressure signal).

  Since the seating noise affects a wide frequency band, it is difficult to separate it by frequency decomposition. However, since the seating noise is generated only at the timing when the valve is seated, it is only necessary to remove this timing from the section subject to frequency decomposition.

  As described above, in order to detect knocking with better accuracy, it is preferable to set a detection interval in a limited manner so as not to include seating noise and to perform frequency decomposition on the detection interval.

  Since knocking occurs in the vicinity of the peak of the in-cylinder pressure, it can be considered that the detection section is set to a predetermined period from the in-cylinder pressure peak. With such a setting, even if the knocking occurrence time fluctuates due to, for example, the variable compression ratio mechanism and aging deterioration, the section in which the in-cylinder pressure should be detected can be set appropriately. Furthermore, by sequentially detecting the peak time of the in-cylinder pressure, it is possible to perform appropriate setting of the detection section in real time.

  However, as described above, when 1) frequency decomposition with a rough frequency characteristic is employed, or 2) when the detection interval is set to be short, the in-cylinder pressure component cannot be removed, and the in-cylinder pressure component is used for knocking detection. May affect.

  Here, the influence of the in-cylinder pressure component in the case of 1) will be considered.

  As a means for realizing time-series frequency decomposition, a discrete wavelet converter using Haar's analyzing wavelet is taken as an example. FIG. 3A shows a configuration diagram of the wavelet transformer having three stages. The wavelet transformer receives an in-cylinder pressure signal P (n) calculated from a sample dP (n) of the in-cylinder pressure change signal obtained by sampling the output signal of the in-cylinder pressure sensor. n indicates the sampling time, and the sampling frequency is Fs.

The first high-frequency subband filter 31 generates the output signal _PH1 according to Equation (1). The first low frequency subband filter 32 generates the output signal P _L1 according to the equation (2). The downward arrow in the formula represents downsampling.

  FIG. 3B is a block diagram showing Expression (1) executed by the first high-frequency subband filter 31. The difference between the current value P (n) of the input signal (here, the in-cylinder pressure signal) and the signal delayed by the delay element 41, that is, the previous value P (n-1) of the input signal is obtained by the differencer 42. calculate. The output of the differentiator 42 is multiplied by (1/2) by the amplifier 43. The output from the amplifier 43 is downsampled by the downsampling circuit 44.

  FIG. 3C is a block diagram showing Expression (2) executed by the first low-frequency subband filter 32. The adder 47 calculates the sum of the current value P (n) of the input signal and the signal delayed by the delay element 46, that is, the previous value P (n-1) of the in-cylinder pressure signal. The output of the adder 47 is multiplied by (1/2) by the amplifier 48. The output from the amplifier 48 is downsampled by a downsampling circuit 49.

  Alternatively, the values multiplied by amplifiers 43 and 48 may use other values. The second and third high-frequency subband filters 33 and 35 also have the same configuration and function as the first high-frequency subband filter 31. The second and third low frequency subband filters 34 and 36 also have the same configuration and function as the first low frequency subband filter 32.

Since the sampling frequency is Fs, the output signals P _H1 and P _L1 of the first high-frequency subband filter 31 and the low-frequency subband filter 32 are generated by Fn (= Fs / 2), and the frequency of the output signal P _H1 The band is Fn / 2 to Fn, and the frequency band of the output signal P _L1 is 0 to Fn / 2. The output signal _{P H2} and _{P L2} of the second high-frequency sub-band filter 33 and the low-frequency sub-band filter 34 is generated by the Fn / 2, the frequency band of the output signal _{P H2} is an Fn / 4~Fn / 2 The frequency band of the output signal P _L2 is 0 to Fn / 4. The output signal _{P H3} and _{P L3} of the third high-frequency sub-band filter 35 and the low-frequency sub-band filter 36 is generated by the Fn / 4, the frequency band of the output signal _{P H3} is an Fn / 8~Fn / 4 The frequency band of the output signal P _L1 is 0 to Fn / 8.

FIG. 4 shows wavelet tiling for the wavelet transformer shown in FIG. When the knocking frequency is about 8 kHz and the in-cylinder pressure sampling frequency Fs is 100 kHz, the signal indicating knocking is included in the output signal _PH3 of the third high-frequency subband filter 35 of the wavelet converter.

  In such wavelet transform, when the high frequency band is decomposed, the update frequency is high (the update cycle is short) and the sampling is fine. Sampling becomes rougher as the frequency band becomes lower, and thus the calculation load can be reduced.

  However, such wavelet transform has a rough frequency characteristic and is easily influenced by other frequencies. If the wavelet transform is used for knocking detection, there is a risk of being affected by in-cylinder pressure.

  FIG. 5 shows such an influence. (A) shows the waveform of a signal indicating knocking (the cylinder pressure signal is zero). In this example, the knocking signal has a substantially sinusoidal waveform of 7.5 to 8.5 kHz. (B) and (c) show an in-cylinder pressure signal obtained from the output of the in-cylinder pressure sensor when knocking occurs. (B) shows the case where the maximum value of the in-cylinder pressure signal is about 3 MPa, and (c) shows the case where the maximum value of the in-cylinder pressure signal is about 6 MPa. Note that the knocking detection section (indicated by two dotted lines) subject to wavelet transform is set to 20 degrees (deg) to 60 degrees after TDC of the compression stroke so as not to be affected by seating noise or the like. ing.

(A1) to (c1) are enlarged views of a signal waveform in the detection section of (a) to (c), and (d) to (f) are graphs with respect to the in-cylinder pressure in the detection section. The output signals obtained by performing the wavelet transform as shown in FIG. Here, since the output signal of the wavelet transform is the signal _PH3 , which is a signal corresponding to the knocking frequency band, it can be considered as a signal indicating the strength of knocking.

  As is clear from these figures, as the in-cylinder pressure increases, the value of the output signal of the wavelet transform also increases (see the transition of peak values in (d) to (f)). This indicates that the in-cylinder pressure component affects the signal representing the knocking strength. As described above, when the wavelet transform having a rough frequency characteristic as shown in FIG. 3 is used, the knocking intensity signal may be influenced by the in-cylinder pressure component.

  Next, the effect of 2) will be described. Even when frequency resolution with fine frequency characteristics is used, if the detection interval is set short, the in-cylinder pressure component may affect knocking detection.

  FIG. 6 shows the state of this influence. (A) shows the waveform of the in-cylinder pressure signal obtained from the output of the in-cylinder pressure sensor when knocking has not occurred. (B) and (c) are the same as (b) and (c) of FIG. 5, and show the in-cylinder pressure signal obtained from the output of the in-cylinder pressure sensor when knocking occurs. (B) shows the case where the maximum value of the in-cylinder pressure signal is about 3 MPa, and (c) shows the case where the maximum value of the in-cylinder pressure signal is about 6 MPa. Note that the knocking detection section (indicated by two dotted lines) subject to wavelet transformation is set to 20 degrees (deg) after the TDC of the compression stroke so as not to be affected by the seating noise and the like, as in FIG. It is set to ~ 60 degrees. (A1)-(c1) is the figure which expanded the waveform of the signal in this detection area of (a)-(c).

  (A) of FIG. 7 shows the result of performing frequency resolution finer than the wavelet transform as shown in FIG. 3 on the signals in the detection sections of (a1) to (c1) of FIG. Reference numerals 51 to 53 indicate the results of the waveforms (a1) to (c1) in FIG. (A1) in FIG. 7 is an enlarged view of the region 50 in (a) (this is a frequency band corresponding to knocking).

  As shown in FIG. 7, the waveform of the frequency decomposition result 51 when knocking does not occur is similar to the waveform of the frequency decomposition results 52 and 53 when knocking occurs. That is, it can be seen that the in-cylinder pressure signal affects the frequency band corresponding to knocking. This is because the in-cylinder pressure frequency band having a longer period than the detection interval becomes uncertain because the detection interval is set short (in this example, the crank angle range is 40 degrees). The occurrence of knocking may be difficult to determine due to the influence of the in-cylinder pressure component.

  The present invention seeks to avoid a decrease in the accuracy of knock detection due to the effects 1) and 2) above.

  FIG. 8 is a block diagram of a knocking detection device according to an embodiment of the present invention. The knocking detection device is implemented in the ECU 1 (FIG. 1). Typically, the function of each block is realized by executing a program stored in the memory of the ECU 1. Alternatively, these functions may be realized by hardware and a combination of software and hardware.

  Knock detection section setting unit 61 sets a section in which knocking should be detected according to the operating state of the engine. Since the knocking detection section is set according to the operating state of the engine, the influence of seating noise and the like can be eliminated.

The in-cylinder pressure component removing unit 62 removes the in-cylinder pressure component in the knocking detection section from the in-cylinder pressure signal obtained through the output of the in-cylinder pressure sensor. The predetermined frequency band extraction unit 63 extracts a signal in a predetermined frequency band including the knocking frequency from the signal from which the in-cylinder pressure component has been removed. Although the predetermined frequency band extraction unit 63 can be realized by any appropriate method, in this embodiment, the predetermined frequency band extraction unit 63 is realized by a wavelet transformer as shown in FIG. By the wavelet transformer, it is possible to extract the signal P _H3 frequency band corresponding to knocking. Since the in-cylinder pressure component is removed, the signal extracted here includes a knocking signal that is not affected by the in-cylinder pressure component. The knocking determination unit 64 determines whether knocking has occurred based on the extracted signal.

  Thus, knocking is determined from the signal from which the in-cylinder pressure component is removed, so that it is possible to avoid the influence of the in-cylinder pressure component on the detection of knocking.

  Each of these blocks will be described in detail below.

  The knocking detection interval setting unit 61 sets the knocking detection interval by any appropriate method. In the first embodiment, a knocking detection section is set according to information indicating engine rotation. As described above, knocking occurs near the in-cylinder pressure peak. That is, it occurs in the vicinity where the piston reaches top dead center (TDC). Therefore, for example, a predetermined crank angle range after TDC of the compression stroke can be set as the knocking detection interval. Information indicating the rotation of the engine can be obtained from, for example, the crank angle sensor 17 (FIG. 1). For example, when the crank angle detected from the output of the crank angle sensor 17 reaches 20 degrees after the TDC of the compression stroke, the knocking detection section is started, and when the crank angle reaches 60 degrees after the TDC, the knocking detection section is ended. . For example, a sensor for detecting TDC may be used as an acquisition source of information indicating the rotation of the engine.

  In the second embodiment, the knocking detection interval setting unit 61 detects the peak of the in-cylinder pressure from the output signal of the in-cylinder pressure sensor, and sets a predetermined period from the peak time as the knocking detection interval.

  FIG. 9A is a block diagram of the knocking detection section setting unit 61. (B) shows the in-cylinder pressure signal P, the output signal Pedg of the differentiator 65, the value of the flag STknock, and the counter value STcnt.

The differentiator 65 receives the output signal dP of the in-cylinder pressure sensor. The differentiator 65 differentiates the in-cylinder pressure change signal dP to generate a signal Pedg as shown in Expression (3). Here, k represents an operation cycle. Since the in-cylinder pressure change signal dP is a primary differential signal of the in-cylinder pressure signal P, the signal Pedg is a secondary differential signal of the in-cylinder pressure signal P. By such differentiation, a signal in which the edge of the in-cylinder pressure signal is emphasized can be obtained, and therefore, the peak of the in-cylinder pressure can be obtained more accurately and efficiently. Alternatively, the in-cylinder pressure signal P may be input to the differentiator 65 and differentiated twice.

  The in-cylinder pressure peak detecting device 66 compares the signal Pedg with a predetermined value PmaxSH every time the signal Pedg is obtained. If the signal Pedg having a value equal to or greater than the predetermined value is obtained, it is determined that the peak of the in-cylinder pressure has been detected, and a value 1 is set in the flag STkock. If the signal Pedg having a value equal to or greater than the predetermined value is not obtained, the flag STknock remains set to zero.

  The knocking detection interval setting device 67 starts measuring the predetermined period ANG3 by the counter in response to the value 1 being set in the peak flag STknock. The counter value STcnt is incremented by a predetermined value every calculation cycle. A section until the counter value STcnt reaches the predetermined period ANG3 is a knocking detection section.

  Alternatively, a signal passage permission device may be provided between the differentiator 65 and the in-cylinder pressure peak detection device 66. Although an impulse signal may be generated due to noise such as seating noise, the signal passage permission device can avoid setting a detection interval in response to such an impulse signal.

  The signal passage permission device determines whether or not the current engine rotation is within a predetermined range. For example, as described in the above-described embodiment, it can be determined whether or not the detected crank angle is within a predetermined range (20 to 60 degrees after the TDC of the compression stroke).

  If the crank angle is within the predetermined range, passage of the signal Pedg to the in-cylinder pressure peak detecting device 66 is permitted. That is, the signal Pedg is subjected to processing by the in-cylinder pressure peak detection device. If the crank angle is not within the predetermined range, the passage of the signal Pedg to the in-cylinder pressure peak detecting device 66 is prohibited.

  Next, the in-cylinder pressure component removing unit 62 will be described. The removal unit 62 can be realized in several forms. These forms are described below.

  In the first embodiment, the in-cylinder pressure component removing unit 62 is realized by a filter. The removal unit 62 performs filtering using the filter every time the in-cylinder pressure signal P is obtained during the knocking detection interval. The in-cylinder pressure signal P can be calculated every time the output signal dP of the in-cylinder pressure sensor is sampled.

  An example of the gain characteristic of the filter is shown in FIG. The filter is an IIR Butterworth filter and a high-pass filter. The cut-off frequency is 4 kHz. The filter order is 2.

A general expression of a filter that realizes the removing unit 62 can be expressed as Expression (4). P indicates an in-cylinder pressure signal, and Paf indicates a filtered in-cylinder pressure signal. A0 to AN and B1 to BN indicate predetermined filter coefficients. k represents an operation cycle. When the filter order is 2, N = 2.

  The frequency component of the in-cylinder pressure depends on the engine speed, but in a general engine (an engine whose upper limit is 6000 rpm), it exists in a low frequency region of about 0 to 2 kHz. Since it is far from the knocking frequency band, the in-cylinder pressure component can be removed with good accuracy even with a filter having a relatively low order.

  Alternatively, the filter may be implemented with a filter of another order so as to remove the above-mentioned in-cylinder pressure frequency component, or may be implemented with another type of filter (for example, a bandpass filter). Good.

  FIG. 11 shows the result of removing the in-cylinder pressure component with a filter having the characteristics shown in FIG. 11A to 11C are the same as the signals shown in FIGS. 5A to 5C. Signal 71 indicates a knocking signal (cylinder pressure signal is zero), and signals 72 and 73 indicate cylinder pressure signals obtained from the output of the cylinder pressure sensor when knocking occurs. The maximum value of the signal 72 is about 3 MPa, and the maximum value of the signal 73 is about 6 MPa.

  (A1) to (c1) are enlarged views of the signal in the detection section after the filter is applied to the signals 71 to 73 of (a) to (c). For (a), after passing through the filter, the signal 71 is left as it is, so the signal 74 is the same as the signal 71 in the detection interval. As for (b) and (c), the signals after passing through the filter are indicated by reference numerals 75 and 76, and these signals are in the vicinity of zero as in (a1). . As shown in these signals after passing through the filter, it is understood that the in-cylinder pressure component is removed by applying the filter.

FIG. 11D shows the result of applying the wavelet transform as shown in FIG. 3 to the signals 74 to 76 that have passed through the filter. As described above, the output signal _{PH3 of} FIG. 3 is obtained as the knocking intensity signal. Output signals 77 to 79 are knocking intensity signals obtained corresponding to signals 74 to 76, respectively. As is apparent from the figure, the peak values of these knocking intensity signals are almost the same. As apparent from comparison with (d) to (f) of FIG. 5, a knocking intensity signal in which the influence of the in-cylinder pressure component is reduced is obtained by performing wavelet transform on the signal after passing through the filter. Can do.

  FIG. 12 shows a block diagram of the in-cylinder pressure component removing unit 62 according to the second embodiment. In this embodiment, the in-cylinder pressure removing unit 62 includes an in-cylinder pressure buffering unit 81, a storage unit 82, an approximate expression calculating unit 83, and a subtracting unit 84. The storage unit 82 can be realized by a memory of the ECU 1 (FIG. 1).

  In order to obtain the knocking frequency band with good accuracy, the output of the in-cylinder pressure sensor 15 may be sampled at a high speed (in the embodiment, 96 kHz). The calculation cycle by the filter is also performed at high speed in accordance with the sampling. The filter calculation at a high cycle may increase the calculation load. However, in the embodiment as shown in FIG. 12, the cylinder pressure component is removed by a simple calculation. Can be suppressed.

  Looking at the waveform of the in-cylinder pressure signal P as shown in FIG. 5 and FIG. 6, the knocking detection interval is short with respect to the length of one combustion cycle (in this example, the crank angle range is 720 degrees). In the detection section, the in-cylinder pressure signal can be approximated to a primary straight line.

  The in-cylinder pressure buffering unit 81 samples the in-cylinder pressure signal P over the knocking detection interval and accumulates it in the storage unit 82. The in-cylinder pressure signal P can be calculated every time the output signal dP of the in-cylinder pressure sensor 15 is sampled. In FIG. 13, the in-cylinder pressure signal P obtained over the detection section is indicated by reference numeral 91.

Of the accumulated in-cylinder pressure signal P, the approximate expression calculation unit 83 uses an initial value Pst at the start point (crank angle: initial angle STANG) of the detection section and a final value Ped at the end point (crank angle: end angle EDANG). , Taken out from the storage unit 82. The approximate calculation unit 83 calculates a primary approximate expression of the in-cylinder pressure as shown in Expression (5). Here, Pcyl_hat indicates an in-cylinder pressure at a certain crank angle ANG estimated by an approximate expression. Ap indicates the slope of the primary line, and Bp indicates the intercept of the primary line.

For example, when STANG = 20 degrees, EDANG = 60 degrees, Pst = 6061 kPa, and Ped = 2049 kPa, Ap, Bp, and Pcyl_hat are as follows. This approximate expression is represented by reference numeral 92.

The subtracting unit 84 subtracts the estimated in-cylinder pressure Pcyl_hat obtained by substituting the crank angle corresponding to the sampling into the equation (5) from the in-cylinder pressure signal P at each sampling, as in the equation (6). The subtraction result Paf indicates a signal from which the in-cylinder pressure component has been removed. The calculation result of Paf is shown in the figure as reference numeral 93.
Paf = P-Pcyl_hat (6)

    FIG. 14 shows the result of applying the wavelet transform as shown in FIG. 3 to the Paf thus obtained. Reference numeral 97 indicates an output of the wavelet transform when the in-cylinder pressure is removed from the knocking signal of FIG. Reference numerals 98 and 99 indicate the outputs of the wavelet transform when the in-cylinder pressure is removed from the in-cylinder pressure signal when the knocking of FIGS. 5B and 5C occurs. Reference numeral 98 denotes a case where the maximum value of the in-cylinder pressure is about 3 MPa, and reference numeral 99 denotes a case where the maximum value of the in-cylinder pressure is about 6 MPa. As apparent from comparison with FIG. 5, the peak values of these knocking intensity signals are substantially the same, and it can be seen that the influence of the in-cylinder pressure component is reduced.

  In this embodiment, the in-cylinder pressure signal in the detection section is approximated by a linear expression. Alternatively, it may be approximated by another approximate expression (for example, a quadratic expression).

As described above, the predetermined frequency band extracting unit 63 can be realized by, for example, a wavelet transformer as shown in FIG. The wavelet transformer extracts a signal (knocking strength signal) in a predetermined frequency band including a knocking frequency from the signal Paf from which the in-cylinder pressure component has been removed by the in-cylinder pressure removing unit 62. In this embodiment, the extracted signal is the output signal _PH3 of the third high-frequency subband filter. Since the operation of the wavelet transformer has been described with reference to FIG. 3, the description thereof is omitted here.

  Next, the knock determination device 64 will be described. The knocking determination device 64 can determine knocking by any appropriate method.

  In the first embodiment, when the time when the knocking strength signal exceeds a predetermined value is equal to or longer than the predetermined time, it is determined that knocking has occurred. According to this method, it is possible to further avoid accidental determination of sudden noise mixed in the knocking frequency band as knocking. Since the time length of the detection section varies depending on the engine speed, the predetermined time is preferably represented by a crank angle or the number of times. However, the predetermined time may be expressed by time.

  In this embodiment, the time during which the knocking intensity signal is equal to or greater than the predetermined value KnkSH is counted by the counter in the detection interval, and if the value of the counter reaches the predetermined number of times TknkSH, it is determined that knocking has occurred.

  In (a) of FIG. 15, since the count number 112 for measuring the time during which the knocking intensity signal 111 in the detection section exceeds the predetermined value KnkSH exceeds the predetermined number TknkSH, it is determined that knocking has occurred. In (b), since the count number 112 does not exceed the predetermined number of times TknkSH, it is determined that there is no knocking.

  In the second embodiment, when the integrated value of the knocking intensity signal is equal to or greater than a predetermined value, it is determined that knocking has occurred. Such a method can also avoid knocking misjudgment caused by sudden noise.

  In FIG. 16A, when the integrated value 114 of the knocking strength signal 113 in the detection section exceeds a predetermined value SknkSH, it is determined that knocking has occurred. In (b), since the integrated value 114 does not exceed the predetermined value SknkSH, it is determined that there is no knocking.

  Further, in an alternative embodiment, the output signal (knocking intensity signal) of the wavelet transformer in the knocking detection section may be examined, and when the knocking intensity signal exceeds a predetermined value, it may be determined that knocking has occurred.

  FIG. 17 shows a flowchart of a process for determining knocking. The process can be realized by the blocks shown in FIG. The process is repeatedly executed at predetermined time intervals.

  In step S1, it is determined whether the current time is within the knocking detection interval. Details are shown in FIG. 18 or FIG.

  If the determination in step S1 is No, this process ends. If the determination of step S1 is Yes, it will progress to step S2 and removal of an in-cylinder pressure component will be performed. Details are shown in FIG. 20 or FIG.

  In step S3, a knocking intensity signal is extracted using wavelet transform. In step S4, it is determined whether knocking has occurred based on the knocking strength signal.

  FIG. 18 shows the process executed in step S1 of FIG. 17 and corresponds to the first embodiment realized by the knocking detection interval setting unit 61 described above. In step S11, it is determined whether or not the crank angle detected at the present time is within a predetermined crank angle range from ANG1 to ANG2. If it is within the predetermined crank angle range, it is determined in step S12 that the current time is within the knocking detection section. If not within the predetermined crank angle range, the process ends.

  FIG. 19 shows another process executed in step S1 of FIG. 17 and corresponds to the second embodiment realized by the knocking detection section setting unit 61 described above.

  In step S21, it is determined whether or not the flag STknock is set to 1. As described above, this flag is a flag that is set to a value of 1 in response to the detection of the peak of the in-cylinder pressure. If the peak is detected, the process proceeds to step S26, and if the peak is not detected, the process proceeds to step S22.

  In step S22, the secondary differential signal Pedg of the in-cylinder pressure is calculated from the difference between the current value dP (k) and the previous value dP (k-1) obtained from the in-cylinder pressure sensor. When calculating the secondary differential signal Pedg from the in-cylinder pressure signal P, the in-cylinder pressure change signal dP is calculated from the difference between the current value and the previous value of the in-cylinder pressure signal P before step S22.

  In step S23, it is determined whether or not the signal Pedg exceeds a predetermined value PmaxSH. If not exceeded, it is determined that the peak is not reached, and in step S24, a flag STknock is set to zero. If it exceeds, it is determined as a peak, the value 1 is set in the flag STknock, and the counter STcnt is reset to zero.

  In step S26, the counter value STcnt is incremented by one. In step S27, it is determined whether the counter value STcnt has reached a predetermined value ANG3. If not reached, it is determined that the current time is within the knocking detection section (S29). If reached, the flag STknock and the counter STcnt are reset to zero (S28).

  FIG. 20 shows the process executed in step S2 of FIG. 17 and corresponds to the first embodiment realized by the in-cylinder pressure removing unit 62 described above. In step S31, the in-cylinder pressure signal P in the detection section is filtered according to the above-described equation (4). By this filtering, a signal Paf from which the in-cylinder pressure component has been removed is obtained. Each time the process of FIG. 20 is performed over a detection interval, a signal Paf is obtained.

  FIG. 21 shows another process executed in step S2 of FIG. 17, and corresponds to the second embodiment realized by the in-cylinder pressure removing unit 62 described above. In step S41, the sample value P (k) obtained by sampling the in-cylinder pressure signal P is buffered. In step S42, n is incremented. n indicates the buffer position. In step S43, it is determined whether n has reached N. If n has not reached N, the process is terminated, and the sampled in-cylinder pressure signal P (k) is stored again in the next buffer position Pbuff (n) in the next calculation cycle. In this way, sample values are stored at 0 to N buffer positions over the detection interval.

  In step S44, the in-cylinder pressure sample value stored in the buffer position of n = 0 is set to the initial value Pst, and the in-cylinder pressure sample value stored in the buffer position of n = N is set to the final value Ped. As described above, these represent the in-cylinder pressure at the start time and end time of the detection section, respectively.

  In step S45, an inclination Ap and an intercept Bp are calculated in order to approximate the in-cylinder pressure signal in the detection section to a linear expression. As described above, the buffer positions 0 and N representing the start and end times of the detection section are used to approximate the linear expression, respectively. However, as described with reference to FIG. 13, the start and end of the detection section are used. The crank angle at the time may be used.

  In step S46, the estimated in-cylinder pressure Pcyl_hat is calculated using the calculated slope Ap and intercept Bp. The estimated in-cylinder pressure Pcyl_hat is calculated corresponding to each buffering time (0 to N).

  In step S47, the calculated estimated in-cylinder pressure Pcyl_hat is subtracted from the corresponding in-cylinder pressure sample value Pbuff to calculate the signal Paf from which the in-cylinder pressure component has been removed. Thus, a signal Paf of 0 to N is obtained over the detection interval.

  FIG. 22 shows a wavelet transform process for extracting a signal of a predetermined frequency band including a frequency component corresponding to knocking, which is executed in step S3 of FIG. This process corresponds to the wavelet transform method described with reference to FIG.

  In the case of the process of FIG. 20, every time the signal Paf is calculated, the process of FIG. 22 can be performed. In the case of the process of FIG. 21, all of the signals Paf (0) to Paf (N) are calculated at the end at the end of the knocking detection interval. Therefore, the process of FIG. 22 is repeated in order for signals Paf (0) to Paf (N).

  Steps S51 to S55 show the processing of the first low-frequency subband filter. In step S51, a shift process for receiving the signal Paf (k) from which the in-cylinder pressure component has been removed is performed. By the process of step S51, the current value Paf (k) of the signal is stored in Pa (k), and the previous value is stored in Pa (k-1).

In step S52, as described with reference to FIG. 3C, the output P _L1 (k) of the first low-frequency subband filter is calculated. Downsampling is performed in steps S53 to S55. If the counter Count is set and the Count value is zero, the Count value is incremented (S55) and the process is terminated. If the value of Count is 1, the value of Count is reset (S54), and the output P _L1 (k) of the filter is passed to the second stage of the wavelet.

Steps S56 to S58 show the process by the second low-frequency subband filter. Step S56 is a shift process similar to step S51. In step S57, the current value P _L1 (k) and the previous value P _L1 (k−1) of the values received from the first stage are obtained as in step S52. Used to generate the output signal P _L2 (n) of the second low-frequency subband filter. In the downsampling process in step S58, the same processes as in steps S53 to S55 are actually performed, but are omitted for simplification.

Steps S59 and S61 show a process by the third high-frequency subband filter. In step S59, a shift process is performed. In step S60, the current value P _L2 (k) and the previous value P _L2 (k−1) received from the second stage are used to output the output signal _{PH3 of} the second high-frequency subband filter. (K) is generated. In step S61, downsampling (same as S53 to S55) is performed.

In step S62, the output signal _{P H3} is set to a knock intensity signal KnockInt.

Assuming that the sampling frequency Fs of the in-cylinder pressure signal is 96 kHz, as described above, the first-stage low-frequency output signal P _L1 is generated at 48 kHz, and the frequency band of the signal is 0 to 24 kHz. The second low frequency output signal P _L2 is generated at 24 kHz, and the frequency band of the signal is 0 to 12 kHz. The third high frequency output signal is generated at 12 kHz, and the frequency band of the signal is 6 to 12 kHz. Thus, the third high-frequency output signal _PH3 is a signal in a frequency band including knocking.

  In this embodiment, a wavelet composed of three stages is used, but this is an example, and for example, a two-stage low-frequency output signal may be used.

  FIG. 23 shows the process executed in step S4 of FIG. 17, which is realized by the aforementioned knock determination unit 64, and corresponds to the first embodiment shown in FIG. In step S71, it is determined whether or not the knocking strength signal KnockInt is greater than or equal to a predetermined value KnkSH. If it is equal to or larger than the predetermined value, the process proceeds to step 72, and the counter Tknk for measuring the time when the knocking intensity signal exceeds the predetermined value is incremented.

  In step S73, it is determined whether or not the counter Tknk is equal to or greater than a predetermined value TknkSH. If it is equal to or greater than the predetermined value, it is determined that knocking has occurred (S74). Otherwise, it is determined that knocking has not occurred (S75).

  FIG. 24 shows another process executed in step S4 of FIG. 17, which is realized by the aforementioned knock determination unit 64, and corresponds to the second embodiment shown in FIG. In step S81, an integrated value Knkint of the knocking intensity signal is calculated. In step S82, it is determined whether or not the integrated value Skinkint is greater than or equal to a predetermined value SknkSH. If it is equal to or greater than the predetermined value, it is determined that knocking has occurred (S83). Otherwise, it is determined that knocking has not occurred (S84).

  In the above embodiment, the engine includes a variable compression ratio mechanism, but the present invention can also be applied to an engine that does not include a variable compression ratio mechanism.

  In the above embodiment, the in-cylinder pressure component removing unit 62 (FIG. 8) has been described as operating with the in-cylinder pressure signal P as an input. Alternatively, the in-cylinder pressure component removing unit 62 may be configured to receive the in-cylinder pressure change signal dP. In this case, in the first embodiment using a filter, for example, a filter having the characteristics described above with reference to FIG. 10 is applied to the in-cylinder pressure change signal dP to obtain a signal from which the in-cylinder pressure component is removed. be able to. Further, in the second embodiment using approximation, the in-cylinder pressure change signal dP in the detection section is approximated to a straight line (as described above, the in-cylinder pressure signal P can be approximated to a linear expression, so the in-cylinder pressure change signal dP Can be approximated to a straight line representing a stationary term (offset)). A signal from which the in-cylinder pressure component is removed can be obtained by subtracting the obtained steady term from the corresponding in-cylinder pressure change signal dP. In either embodiment, the signal from which the in-cylinder pressure component has been removed can be directly input to the wavelet transformer.

  The present invention is applicable to a general-purpose internal combustion engine (such as an outboard motor).

The figure which shows schematically the engine and its control apparatus according to one Example of this invention. The figure which shows an example of the result of having analyzed the frequency of the output signal of a cylinder pressure sensor. 1 is a block diagram of a wavelet transformer according to one embodiment of the present invention. FIG. FIG. 3 is a diagram illustrating wavelet tiling according to one embodiment of the present invention. The figure for demonstrating the influence which it has on the knocking detection of a cylinder pressure component. The figure for demonstrating the influence which it has on the knocking detection of a cylinder pressure component. The figure for demonstrating the influence which it has on the knocking detection of a cylinder pressure component. 1 is a block diagram of a knocking detection device according to one embodiment of the present invention. The figure for demonstrating the structure of the knocking detection area setting part and its method according to one Example of this invention. The figure which shows the characteristic of the filter used by the cylinder pressure removal part according to one Example of this invention. The figure for demonstrating the effect of the removal of the in-cylinder pressure component by a filter according to one Example of this invention. The block diagram which shows the structure of the cylinder pressure removal part according to the other Example of this invention. The figure for demonstrating the method of removing the in-cylinder pressure component according to the other Example of this invention. The figure for demonstrating the effect of the removal of a cylinder pressure component according to the other Example of this invention. The figure for demonstrating the knock determination method according to one Example of this invention. The figure for demonstrating the knock determination method according to the other Example of this invention. 4 is a flowchart of a process for determining knocking, according to one embodiment of the present invention. The flowchart of a knocking detection area setting process according to one Example of this invention. The flowchart of a knocking detection area setting process according to another embodiment of the present invention. The flowchart of the in-cylinder pressure removal process according to one Example of this invention. The flowchart of the in-cylinder pressure removal process according to the other Example of this invention. 4 is a flowchart of a wavelet transform process according to one embodiment of the present invention. 5 is a flowchart of a knock determination process according to an embodiment of the present invention. 7 is a flowchart of a knock determination process according to another embodiment of the present invention.

Explanation of symbols

1 ECU
2 Engine 8 Combustion chamber 15 In-cylinder pressure sensor 17 Crank angle sensor

Claims (10)

In-cylinder pressure detecting means for detecting the in-cylinder pressure of the engine and outputting a signal corresponding to the in-cylinder pressure;
Removing means for removing an in-cylinder pressure component in the knocking detection section from an output signal of the in-cylinder pressure detecting means;
Extraction means for extracting a signal in a predetermined frequency band including a knocking frequency from the signal from which the in-cylinder pressure component has been removed;
A knocking detection apparatus comprising: a determination unit that determines occurrence of knocking from the extracted signal. And a means for setting the knocking detection section based on information indicating rotation of the engine.
The knock detection device according to claim 1. Furthermore, it comprises means for detecting the peak of the in-cylinder pressure from the output signal of the in-cylinder pressure detecting means, and setting the knocking detection section in a predetermined period from the time of the peak.
The knock detection device according to claim 1. The removing means removes the in-cylinder pressure component by a filter configured to cut a low-frequency component including the in-cylinder pressure component.
The apparatus according to claim 1. The removing means calculates an approximate expression for approximating the in-cylinder pressure from the output signal of the in-cylinder pressure detecting means in the knocking detection section, and the cylinder calculated by the approximate expression from the output signal of the in-cylinder pressure detecting means. The in-cylinder pressure component is removed by subtracting the internal pressure.
The apparatus according to claim 1. The approximate expression is a linear expression, and the removing means obtains the linear expression from the output signal of the in-cylinder pressure detecting means at the starting point and the ending point of the knocking detection section to obtain the slope and intercept of the linear expression. calculate,
The apparatus according to claim 5. The output signal of the in-cylinder pressure detecting means is a signal representing a differential value of the in-cylinder pressure.
The apparatus according to claim 1. The determination means determines that knocking occurs when the time that the signal extracted by the extraction means is equal to or greater than a predetermined value continues for a predetermined time or longer.
The apparatus according to claim 1. The determination means integrates the signal extracted by the extraction means over the knocking detection section, and determines that knocking occurs if the integrated value exceeds a predetermined value.
The apparatus according to claim 1. The engine is an engine having a variable compression ratio mechanism capable of changing the compression ratio in the combustion chamber.
Apparatus according to any of claims 1 to 9.

Images