JPH11315750A - Maximum value detection method for sensor signal, sensor signal processing device, and knock detection method for internal combustion engine - Google Patents

Abstract

PROBLEM TO BE SOLVED: To accurately detect the maximum value of oscillation with respect to the vibration center level of a sensor signal which is a vibration wave, without influence from noise and regardless of the length of detection time. SOLUTION: An engine control ECU, which inputs a sensor signal from a knock sensor between a knock judgment interval and which detects the maximum value of oscillation with respect to the vibration center level (reference level) of that signal, integrates the oscillation with respect to the reference value of the sensor signal every half wave time T minute of the sensor signal while calculating the average value AVR in which each of those integral values are divided by the continuous time of the half wave time T, respectively. Then, a maximum average value AVRmax is selected from among each average value AVR and the maximum value of oscillation is calculated from those maximum average values AVRmax. This ECU enables accurate detection of the above-mentioned maximum value of oscillation without being influenced from noise and regardless of the length of the knock judgment interval because of the integration and averaging every half wave time T minute.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a sensor signal output as a vibration wave from a sensor for a predetermined detection period, and detects a maximum value of a fluctuation width of the sensor signal with respect to a vibration center level in the detection period. Related to technology.

[0002]

2. Description of the Related Art Conventionally, for example, in a vehicle, a knock sensor has been used to detect knock (knock) generated in an internal combustion engine. As this type of knock sensor, as described in Japanese Patent Application Laid-Open No. 9-329052, a vibration-type knock sensor that outputs a vibration wave corresponding to vibration transmitted by a knock transmitted to a cylinder block of an internal combustion engine as a sensor signal is used. And JP-A-6-159129.
As described in Japanese Patent Application Laid-Open Publication No. H10-209, there is a sensor signal that outputs an oscillating wave corresponding to an ionic current flowing in a cylinder of an internal combustion engine as a sensor signal.

When such a knock sensor is used, if the maximum value of the swing width in the positive direction and / or the negative direction with respect to the vibration center level of the sensor signal is known, the presence or absence of knock is determined from the maximum value. Knock strength can be determined. For this reason, in the electronic control device that controls the internal combustion engine, in a knock determination section that is set as a period in which knock is considered to occur in the internal combustion engine, for example, in the positive direction with respect to the vibration center level of the sensor signal output from the knock sensor. Hold the swing width of the peak,
The presence / absence of knock and the strength of knock are determined using the peak hold value at the end of the knock determination section. That is, in the case of this example, the peak hold value at the end of the knock determination section is the maximum level in the knock determination section of the sensor signal, and the presence or absence of knock is determined based on the value.

However, if a large spike-like noise is superimposed on the sensor signal and the noise is peak-held, the peak hold value at the end of the knock determination section becomes, for example, the true maximum of the sensor signal. It becomes larger than the level, and as a result, the presence / absence of knock or the like is erroneously determined.

Therefore, as a technique for removing the influence of such noise, for example, Japanese Unexamined Patent Publication No. 6-159129 discloses an absolute value of a sensor signal level (specifically, a positive direction and a negative It describes that the swing width in both directions is integrated over a knock determination section, and the presence or absence of knock is determined based on an integrated value at the end of the knock determination section. In other words, when the sensor signal is integrated, the ratio of the noise component in the integrated value becomes smaller, so that it is possible to suppress the influence of the noise when determining the presence or absence of knock or the like.

[0006]

However, as described in JP-A-6-159129, when integration is performed over the entire knock determination section, the total of the sensor signal in the knock determination section is reduced. However, the maximum value (maximum peak value) of the true swing width with respect to the vibration center level of the sensor signal cannot be detected. As a result, the presence or absence of knock and the strength of knock cannot be determined in more detail.

In general, a knock determination section as a detection period for inputting a sensor signal from a knock sensor is:
As illustrated in FIG. 13, the rotation angle position of the crankshaft of the internal combustion engine is 10 ° ahead of the compression stroke top dead center (hereinafter, referred to as TDC) of each cylinder (hereinafter, referred to as ATDC).
10 ° CA), a period until the rotational angle position further advances by 80 ° (that is, a rotational angle position advanced by 90 ° from TDC, hereinafter referred to as ATDC 90 ° CA), and so on. It is set based on the rotation angle position of the crankshaft. In FIG. 13, (a)
Shows a case where the rotation speed of the internal combustion engine is, for example, 1000 [rpm], and (b) shows a case where the rotation speed of the internal combustion engine is, for example, 3 [rpm].
000 [rpm].

Therefore, as can be seen from a comparison between FIGS. 13A and 13B, the length of the knock determination section (AT
(Period from DC10 ° CA to ATDC90 ° CA)
Changes according to the rotation speed of the internal combustion engine, and becomes longer as the rotation speed of the internal combustion engine is lower. For this reason, in the technique described in JP-A-6-159129, even when the maximum value of the swing width with respect to the vibration center level of the sensor signal is the same, the longer the knock determination section, the larger the integral value of the sensor signal becomes. As a result, it becomes impossible to accurately determine the presence or absence of knock or the like.

SUMMARY OF THE INVENTION The present invention has been made in view of the above-mentioned problems, and it has been proposed that a maximum value of a swing width in both or one of a positive direction and a negative direction with respect to a vibration center level of a sensor signal be determined.
A first object is to enable accurate detection without being affected by noise and regardless of the length of the detection period. It is a second object of the present invention to accurately detect knock of the internal combustion engine without being affected by noise and regardless of the length of the detection period.

[0010]

Means for Solving the Problems and Effects of the Invention A method for detecting the maximum value of a sensor signal according to the present invention, which has been made to achieve the first object, comprises the steps of: Is input during the detection period, and the maximum value of the swing width in the positive direction and / or the negative direction with respect to the vibration center level of the sensor signal in the detection period is detected.

According to the maximum value detecting method of the present invention, the swing width of the sensor signal with respect to the vibration center level is determined by the half-wave period in which the sensor signal level is on the positive side of the vibration center level and the sensor signal level by the vibration center. Integrate every minute for both of the half-wave periods on the negative side of the level, or for each one of the two half-wave periods, and divide each integrated value by the duration of the half-wave period. The average value is calculated. Therefore, each of the average values calculated in this manner is an average value of the swing width with respect to the vibration center level of the sensor signal for each half-wave period.

Further, in the maximum value detecting method of the present invention, the maximum value of the swing width is obtained from the maximum value among the calculated average values. In addition, if the maximum value of the swing width in both the positive direction and the negative direction with respect to the vibration center level of the sensor signal is detected, the swing width with respect to the vibration center level of the sensor signal is determined. The integration is performed for each of both the half-wave period on the positive side of the level and the half-wave period on the negative side of the level of the sensor signal from the oscillation center level (that is, for each half-wave period of the sensor signal) Just do it. Also, if the maximum value of the swing width in the positive direction with respect to the vibration center level of the sensor signal is detected, the swing width with respect to the vibration center level of the sensor signal is
The integration may be performed for each half-wave period in which the level of the sensor signal is on the positive side of the vibration center level. Further, if the maximum value of the swing width in the negative direction with respect to the vibration center level of the sensor signal is detected, the swing width with respect to the vibration center level of the sensor signal is set to a value on the negative side of the vibration center level with respect to the sensor signal. In this case, the integration may be performed every half-wave period.

Here, the principle of detecting the maximum value in the present invention will be described in detail. First, the waveform of the sensor signal for a half-wave period is, as shown in FIG. Assuming that the waveform has a waveform, the average value Va of the vibration width with respect to the vibration center level of the sensor signal for a half-wave period is
Equation 1 below is obtained, and the maximum swing width Vp (hereinafter, referred to as a peak value in a half-wave period) from the vibration center level in the half-wave period of the sensor signal can be obtained by the following equation 2.

[0014]

(Equation 1)

[0015]

Vp = (π / 2) × Va (2) Although not shown, for example, if the waveform of a half-wave period of a sensor signal is a triangular wave, the relationship with respect to the vibration center level of the sensor signal is obtained. Average value V for half-wave period of swing width
If a is doubled, the peak value Vp (= 2 ×
Va) can be determined.

That is, if the average value Va of the swing width with respect to the vibration center level of the sensor signal for the half-wave period is known, the peak value Vp of the half-wave period can be calculated. Therefore, in the maximum value detection method of the present invention, the swing width of the sensor signal with respect to the vibration center level is integrated for each half-wave period in which the level of the sensor signal is closer to the detection target direction than the vibration center level as described above. Then, by dividing each of the integrated values by the duration of the half-wave period (that is, the integration time), the average value of the swing width with respect to the vibration center level of the sensor signal for each half-wave period is calculated. From the maximum value among the average values, the maximum value of the fluctuation width with respect to the vibration center level of the sensor signal during the detection period is obtained.

In order to determine the maximum value of the swing width from the maximum value among the average values, (α): select the maximum value among the average values, and calculate a half value from the selected average value. The peak value Vp during the wave period is calculated, and the calculated peak value V
(p): calculating a peak value Vp in a half-wave period for each average value, and determining the largest one of the calculated peak values Vp, Both of these methods are considered. However, it is more advantageous to employ the method (α) because it takes only one time to calculate the peak value Vp in the half-wave period from the average value.

In such a maximum value detecting method of the present invention,
Since the fluctuation width of the sensor signal with respect to the vibration center level is integrated and averaged for each half-wave period, even if noise is superimposed on the sensor signal, the influence of the noise can be extremely suppressed. In addition, since the swing width of the sensor signal with respect to the vibration center level is integrated not for the detection period but for each half-wave period, regardless of the length of the detection period, the sensor signal in the detection period is not affected. The maximum value of the swing width in the positive direction and / or the negative direction with respect to the vibration center level can be accurately detected.

Therefore, according to the maximum value detecting method of the present invention, the maximum value of the swing width in the positive direction and / or the negative direction with respect to the vibration center level of the sensor signal can be determined without being affected by noise and Accurate detection is possible regardless of the length of the detection period. And in particular, as described in claim 2,
When the sensor is a knock sensor for detecting knock generated in the internal combustion engine, there is a high possibility that noise is superimposed on the sensor signal, and the length of the knock determination section as the detection period is set as described above. Will change,
According to the maximum value detection method of the present invention, the maximum value of the swing width in the positive direction and / or the negative direction with respect to the oscillation center level of the sensor signal is determined without affecting the noise and the length of the detection period. Therefore, the presence or absence of knock of the internal combustion engine and the strength of knock can be accurately determined.

However, the maximum value detection method of the present invention is not limited to the case where the sensor signal from the knock sensor is to be detected, but can be applied to any sensor signal as a vibration wave. On the other hand, the sensor signal processing device of the present invention according to claim 3 inputs a sensor signal output as a vibration wave from a sensor for a predetermined detection period, and implements the above-described maximum value detecting method of the present invention. Thus, the maximum value of the swing width in the positive direction and / or the negative direction with respect to the vibration center level of the sensor signal during the detection period is detected.

That is, in the sensor signal processing device of the present invention,
The half-wave period specifying means includes a half-wave period in which the level of the sensor signal is on the positive side of the vibration center level (hereinafter, when this half-wave period is particularly distinguished, a positive half-wave period) and a sensor signal (Hereinafter referred to as a negative half-wave period when this half-wave period is particularly distinguished) is specified, and the integration means Then, the swing width of the sensor signal with respect to the vibration center level is integrated for each half-wave period specified by the half-wave period specifying means.

Then, the average value calculating means calculates an average value obtained by dividing the integral value by the integrating means by the duration of the half-wave period, and the maximum value calculating means calculates the above-mentioned (α) or (β).
The maximum value of the swing width is calculated from the maximum value among the average values calculated by the average value calculation means by the above method.

In such a sensor signal processing apparatus of the present invention, if the maximum value of the swing width in both the positive direction and the negative direction with respect to the vibration center level of the sensor signal is detected, the half-wave period specifying means However, the configuration may be such that both the positive half-wave period and the negative half-wave period are specified. Then, the integrating means integrates the swing width of the sensor signal with respect to the vibration center level for each half-wave period of the sensor signal, and the average value calculating means integrates each half of the swing width of the sensor signal with respect to the vibration center level. An average value is calculated for each wave period. Then, the maximum value calculating means, from the largest one of the average values calculated by the average value calculating means,
The maximum value of the swing width in both the positive direction and the negative direction with respect to the vibration center level of the sensor signal is calculated.

Further, if the maximum value of the swing width in the positive direction with respect to the vibration center level of the sensor signal is detected, the half-wave period specifying means may be configured to specify only the positive half-wave period. good. Then, the integrating means integrates the swing width of the sensor signal with respect to the vibration center level for each positive half-wave period, and the average value calculating means integrates the swing width of the sensor signal with respect to the vibration center level with the positive half-wave. An average value for each period is calculated. Then, the maximum value calculating means calculates the maximum value of the swing width in the positive direction with respect to the vibration center level of the sensor signal from the largest one of the average values calculated by the average value calculating means.

Still further, if the maximum value of the swing width in the negative direction with respect to the vibration center level of the sensor signal is detected, the half-wave period specifying means may be configured to specify only the negative half-wave period. Good. Then, the integrating means integrates the fluctuation width of the sensor signal with respect to the vibration center level for each negative half-wave period, and by the average value calculating means,
An average value of the swing width with respect to the vibration center level of the sensor signal for each negative half-wave period is calculated. Then, the maximum value calculating means calculates the maximum value of the swing width in the negative direction with respect to the vibration center level of the sensor signal from the largest one of the average values calculated by the average value calculating means.

According to such a sensor signal processing device of the present invention, the effect of the above-described maximum value detecting method of the present invention can be obtained.
That is, the maximum value of the swing width in the positive direction and / or the negative direction with respect to the vibration center level of the sensor signal can be accurately detected without being affected by noise and regardless of the length of the detection period. The effect can be obtained.

As a value obtained by dividing the integrated value by the average value calculating means by the integrating means (ie, the duration of the half-wave period), a known half cycle time of the sensor signal may be used.
According to this configuration, it is possible to more accurately detect the maximum value of the fluctuation width of the sensor signal with respect to the vibration center level.

That is, in the sensor signal processing device according to the fourth aspect, the time detecting means detects the duration of the half-wave period specified by the half-wave period specifying means. Then, the average value calculating means calculates the average value by dividing the integrated value by the integrating means by the duration detected by the time detecting means.

According to the sensor signal processing device of the fourth aspect, even if the period of the sensor signal (especially, the actual duration of the half-wave period) changes, the average value is calculated by the average value calculating means. Each average value is always an accurate value, so that the maximum value of the fluctuation width of the sensor signal with respect to the vibration center level can be detected more accurately.

On the other hand, the half-wave period specifying means may be configured to specify the half-wave period of the sensor signal as a period from when the level of the sensor signal crosses the vibration center level until it crosses the vibration center level again. it can. This makes it possible to simplify the processing and circuit for specifying the half-wave period.

However, when such a configuration is applied to the sensor signal processing device according to claim 4, the noise superimposed on the sensor signal causes the level of the sensor signal to be substantially equal to the oscillation center level even during a half-wave period. , The duration of the half-wave period specified by the half-wave period specifying unit becomes short, and accordingly, the duration of the half-wave period detected by the time detecting unit also becomes short. Then, there is a possibility that the average value calculated by the average value calculation means becomes excessively large, and eventually, the maximum value of the swing width calculated by the maximum value calculation means becomes larger than the true maximum value. .

Therefore, in order to solve such a problem, it is sufficient to provide a determining means according to the fifth aspect. That is,
The determining means determines whether or not the duration detected by the time detecting means is within a predetermined setting range. If the duration is not within the setting range, the determination is not within the setting range. The integrated value by the integrating means corresponding to the half-wave period of the duration is discarded.

Then, the average value of the discarded integrated value is no longer calculated by the average value calculating means, and is excluded from the target for calculating the maximum value of the swing width of the sensor signal with respect to the vibration center level. The Rukoto.
Therefore, according to the sensor signal processing device of the fifth aspect including such a determination unit, the influence of noise is further reduced, and the maximum value of the fluctuation width of the sensor signal with respect to the vibration center level is detected more accurately. Will be able to

According to a sixth aspect of the present invention, there is provided the sensor signal processing apparatus according to the third or fourth aspect, wherein the difference between the level of the sensor signal and the vibration center level is detected at regular time intervals. The detecting means is provided, and the integrating means is configured to integrate the amplitude of the sensor signal with respect to the vibration center level by sequentially adding the absolute values of the values detected by the level detecting means. That is, claim 6
In the sensor signal processing device described in (1), the fluctuation width of the sensor signal with respect to the vibration center level is integrated by a sampling method at predetermined time intervals by digital signal processing.

[0035] In particular, the sensor signal processing device according to claim 6 includes a correction means, and the correction means includes:
When only the second detection value is equal to or less than 0 among the three detection values successively detected by the level detection means, or when only the second detection value is equal to or more than 0, the second The absolute value of the average value of the first and third detection values of the three detection values is added to the integration means in place of the absolute value of the detection value of.

According to the sensor signal processing device of the sixth aspect, the noise is superimposed on the sensor signal, and only the second detection value among the three consecutive detection values by the level detection means is 0 or less. If the second detection value is equal to or greater than 0, instead of the second detection value, the average value of the two immediately preceding and succeeding detection values is used. Is regarded as the original value of the second detection value, and is used for integration.

According to the sensor signal processing device of the sixth aspect, the influence of noise is further reduced, and the maximum value of the fluctuation width of the sensor signal with respect to the vibration center level can be detected more accurately. become. And according to the sensor signal processing device of the said Claims 3-6,
As described in claim 7, when the sensor is a knock sensor for detecting a knock generated in the internal combustion engine,
Since the maximum value of the swing width in the positive direction and / or the negative direction with respect to the vibration center level of the sensor signal can be accurately detected without being affected by noise and regardless of the length of the detection period, It is possible to accurately determine the presence or absence of knocking of the engine and the strength of knocking.

On the other hand, a knock detection method for an internal combustion engine according to the present invention according to claim 8 which has been made to achieve the second object, provides a knock sensor for detecting a knock generated in the internal combustion engine. Is input during a predetermined detection period, and the knock of the internal combustion engine is detected from the sensor signal during the detection period.

Here, in the knock detection method of the present invention, similarly to the above-described maximum value detection method of the present invention, the swing width of the sensor signal with respect to the vibration center level is determined by setting the sensor signal level to a more positive side than the vibration center level. In each of both the half-wave period and the half-wave period in which the level of the sensor signal is on the negative side of the oscillation center level, or integrating each half of the two half-wave periods, Average value obtained by dividing each integrated value by the duration of the half-wave period (that is, the average value of the swing width with respect to the vibration center level of the sensor signal for each half-wave period)
Is calculated. Then, the knock of the internal combustion engine is detected from the largest one of the calculated average values.

When the knock is detected, for example, the maximum value of the fluctuation width with respect to the vibration center level of the sensor signal is obtained from the maximum value among the average values in the same manner as in the above-described maximum value detection method. If the maximum value is larger than a predetermined determination value, it may be determined that knock has occurred in the internal combustion engine, or the maximum value among the average values may be larger than the predetermined determination value. If it is larger, it may be determined that knock has occurred in the internal combustion engine.

According to the knock detection method of the present invention, the swing width of the sensor signal with respect to the vibration center level is not integrated over the detection period, as in the above-described maximum value detection method. Since integration and averaging are performed for each half-wave period, knock of the internal combustion engine can be accurately detected without being affected by noise and regardless of the length of the detection period.

[0042]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An electronic control unit for controlling a four-cylinder engine (internal combustion engine) mounted on a vehicle will be described below with reference to the drawings. First Embodiment First, FIG. 2 is a block diagram showing a configuration of an electronic control unit (hereinafter, referred to as an ECU) 1 according to a first embodiment.

As shown in FIG. 2, the ECU 1 inputs a sensor signal (hereinafter referred to as a knock sensor signal) output as a vibration wave from a knock sensor 3 mounted on a cylinder block of the engine, and receives the knock sensor signal. Of the frequency components specific to knock (6 kHz to
Bandpass filter (BP) that passes only 10 kHz
F) 5 and a pulse signal or a switch signal from a rotation angle sensor 7 that outputs a pulse signal each time the crankshaft of the engine rotates a predetermined angle, an accelerator switch 9 that is turned on when an accelerator pedal is depressed, and the like, An input circuit 11 for shaping and outputting a signal is provided, and a drive circuit 17 for driving an actuator such as an ignition system device 13 or an injector 15 of the engine.

The ECU 1 further includes a knock sensor signal output from the band-pass filter 5, an intake air amount sensor 19 for detecting an intake air amount of the engine, and a water temperature sensor 21 for detecting a cooling water temperature of the engine. A multiplexer (M) that receives analog sensor signals output from the like and selectively switches and outputs these signals.
PX) 23, an A / D converter 25 for converting the level of an analog signal output from the multiplexer 23 into digital data (A / D conversion), and the multiplexer 23
And the A / D converter 25 to control the multiplexer 23
CPU 27 that performs signal processing for converting the level of a sensor signal input to the CPU into digital data and obtaining the digital data
And a control process for controlling the engine based on the digital data acquired by the first CPU 27, the pulse signal and the switch signal output from the input circuit 11, and the like. 2nd CPU which outputs drive signal for driving 15 etc.
29, a ROM 31 storing a program to be executed by the first CPU 27 and the second CPU 29, and a RAM 3 for temporarily storing calculation results of the first CPU 27 and the second CPU 29 and the like.
3 is provided.

In this embodiment, the multiplexer 23, the A / D converter 25, the first CPU 27, the second C
The PU 29, the ROM 31, and the RAM 33 are mounted on the ECU 1 as one microcomputer (microcomputer) 35. As shown in the lower part of FIG. 6 and the upper part of FIG. 7 in which P1 in FIG. 6 is enlarged, the knock sensor signal ranges from 0 V to 5 V with a reference value of 2.5 V as a vibration center level. Vibrates at 6kHz-10
It is a vibration wave of kHz and almost a SIN wave.

The EC of this embodiment configured as described above
In U1, the second CPU 29 outputs the signals from the rotation angle sensor 7, the accelerator switch 9, and the like, which are input via the input circuit 11, and the sensor signals from the intake air amount sensor 19, the water temperature sensor 21, and the like acquired by the first CPU 27. The optimum fuel injection amount to the engine, the ignition timing, and the like are calculated based on the level and the ignition system equipment 1
The engine is controlled by driving the fuel injector 3, the injector 15, and the like.

In particular, in the ECU 1 of the present embodiment,
The first CPU 27 and the second CPU 29 cooperate to form the maximum value of the swing width in both the positive direction and the negative direction with respect to the vibration center level (reference value = 2.5 V) of the knock sensor signal (in other words, the knock sensor signal). Of the maximum level and minimum level of the level that is farthest from the vibration center level)
Is detected as the maximum peak value P / Hmax of the knock sensor signal. Then, the second CPU 29 determines whether or not knock has occurred in the engine based on the maximum peak value P / Hmax. If it is determined that knock has occurred, the ignition timing of the engine is corrected to the retard side. Do
Knocking is suppressed.

Therefore, the first CPU 27 and the second CPU 27 will now be described in order to determine whether or not knock has occurred in the engine.
Processing executed by the PU 29 will be described. First,
The second CPU 29 sequentially detects the rotation angle position of the crankshaft of the engine based on the pulse signal from the rotation angle sensor 7, and determines the rotation angle position of the crankshaft as ATD.
C10 ° CA (Rotation angle position advanced 10 ° from TDC)
, The knock determination section start process shown in FIG. 3 is executed.
When the rotation angle reaches CA (a rotation angle position advanced by 90 degrees from TDC), the knock determination section stop processing of FIG. 5 is executed.

On the other hand, the first CPU 27
Is a period from execution of the knock determination section start processing of FIG. 3 to execution of the knock determination section stop processing of FIG. 5, and the rotational angle position of the crankshaft of the engine is shown in FIG.
ATDC10 ° CA to ATDC90 ° CA as shown in
In the knock determination section in the range up to
The process is repeatedly executed every μs. And the first CPU 2
7 inputs the knock sensor signal from the multiplexer 23 to the A / D converter 25 every time the processing is executed every 2 μs, and sets the level of the knock sensor signal to A / D.
The converter 25 performs A / D conversion, and performs processing for detecting the maximum peak value P / Hmax of the knock sensor signal using the A / D converted value. In addition, ATDC10 °
The period from CA to ATDC 90 ° CA is defined as a knock determination section because knock occurs during that period.

Next, the processing executed by the first CPU 27 and the second CPU 29 will be specifically described with reference to the flowcharts of FIGS. In the following description,
Each of “NEWAD” and “OLDAD” is the first CP
U27 sets the knock sensor signal level to 2 μs among the data updated and stored in the RAM 33 by processing every 2 μs.
This is a detection value (see S210 in FIG. 4) obtained by subtracting a reference value (= 2.5 V) as a vibration center level from each A / D conversion value obtained by A / D conversion for each. “NEWAD” is a value detected by the first CPU 27 in the current process every 2 μs, and “OLDAD” is a value detected by the first CPU 27 in the last 2 μs.
This is a value detected in each process. “AVRmax” is
Among the data updated and stored in the RAM 33 by the first CPU 27 every 2 μs, the average value AVR of the amplitude for each half-wave period T with respect to the reference value of the knock sensor signal (see FIG. 7).
(See lower section below).

First, FIG. 3 shows that the second CPU 29
It is a flowchart showing a knock determination section start process executed at the timing of 0 ° CA. As shown in FIG. 3, when the second CPU 29 starts execution of the knock determination section start process, first, in step (hereinafter simply referred to as “S”) 100, an initial value for setting data in the RAM 33 to an initial value is set. Perform the conversion process. By this initialization processing, at least each value of “AVRmax” and “NEWAD” and the integrated value storage area S set in the RAM 33 are stored.
The values of the UM and the counter N are initialized to zero.

Then, the second CPU 29 proceeds to S110.
After allowing the first CPU 27 to execute the processing every 2 μs, the knock determination section start processing ends. Then, thereafter, the first CPU 27 repeatedly executes the processing of FIG. 4 every 2 μs every 2 μs as its name suggests.

That is, as shown in FIG.
Starts execution of the process every 2 μs, first, in S200, the previous “NEWAD” is stored again as “OLDAD”, and in S210, a command is given to the multiplexer 23 to knock the A / D converter 25. A sensor signal is input, and the level of the knock sensor signal is changed to A /
A / D conversion is performed by the D converter 25, and a value obtained by subtracting a reference value (= 2.5 V) from the current A / D conversion value is referred to as “NEWA”.
D ”.

Then, in subsequent S220, it is determined whether or not the relationship of the following equation 3 is established.

[0055]

"OLDAD" ≤0 <"NEWAD" (Equation 3) That is, as shown in P2 in FIG. 7 and FIG. At the zero crossing point where the value changes from the negative side smaller than the reference value to the positive side larger than the reference value and crosses the reference value from the negative side to the positive side, the relationship of the above equation 3 is established.
Therefore, in S220, two consecutive detection values (“OL
DAD ”,“ NEWAD ”) is determined to be in a changing state that satisfies the relationship of Expression 3, thereby determining whether the level of the knock sensor signal has crossed the reference value from the negative side to the positive side. I have.

If it is determined in S220 that the relationship of Expression 3 is not established, the process proceeds to S230, and it is determined whether or not the relationship of Expression 4 below is established.

[0057]

(OLDAD) ≧ 0> “NEWAD” (4) That is, as shown in FIG. 7, the portion of P3 and the level of the knock sensor signal as shown in FIG. At the zero crossing point where the value changes from the positive side larger than the reference value to the negative side smaller than the reference value and crosses the reference value from the positive side to the negative side, the relationship of the above equation 4 holds.
Therefore, in S230, two consecutive detection values (“OL
DAD ”,“ NEWAD ”) is determined to be in a change state that satisfies the relationship of Expression 4, thereby determining whether the level of the knock sensor signal has crossed the reference value from the positive side to the negative side. I have.

Here, when it is determined in S230 that the relationship of Equation 4 is not established, that is, when the level of the knock sensor signal does not cross the reference value, the positive and negative polarities of "OLDAD" and "NEWAD" are determined. Are the same, then S
Proceeding to 240, the absolute value of "NEWAD" (that is, the amplitude of the knock sensor signal detected this time relative to the reference value) is added to the value stored in the integral value storage area SUM, and the value after the addition (= | NEWAD | + SUM) is stored again in the integrated value storage area SUM. Then, in S250,
After incrementing the value of the counter N for counting the number of additions of the absolute value of “NEWAD” by one,
The process is terminated once every μs.

On the other hand, if it is determined in step S220 that the relationship of expression 3 is established, or if it is determined in step S230 that the relationship of expression 4 is established, the level of the knock sensor signal is used as a reference. It is determined that the value has been crossed, and S260
Then, the value stored in the integrated value storage area SUM at that time is divided by the value of the counter N to obtain the average value A.
Calculate VR (= SUM / N).

Next, in subsequent S270, it is determined whether or not the average value AVR calculated in S260 is larger than "AVRmax". If the average value AVR is not larger than "AVRmax" (S270: NO), The process directly proceeds to S290, but if the average value AVR is larger than “AVRmax” (S270: YES), the average value AVR is stored again as “AVRmax” in the next S280, and then S2
Go to 90.

Then, in S290, the current "NEWA"
The absolute value of “D” is stored as an initial value in the integral value storage area SUM, and in S300, the value of the counter N is initialized to 1. Then, thereafter, the process is terminated once every 2 μs. That is, in the processing every 2 μs in FIG.
In the processing of S210, the level of the knock sensor signal is A / D-converted every 2 μs to detect the difference between the A / D-converted value and the reference value. Two consecutive items ("NEWAD",
"OLDAD").

Then, in the processing every 2 μs in FIG.
In the processing of 220 and S230, “NEWAD” and “O
LDAD ", from the time when it is determined whether or not the level of the knock sensor signal has crossed the reference value to the time when it is determined that the level of the knock sensor signal has crossed the reference value, until the determination is made that the level has crossed the reference value again (That is, a period from the time when the affirmative determination is made in S220 to the time when affirmative determination is made in S230, or from the time when the affirmative determination is made in S230 to S220
Is determined as a half-wave period of the knock sensor signal.

It should be noted that from the time when the affirmative determination is made in S220, S2
The period until the affirmative determination in 30 can be regarded as a positive half-wave period in which the level of the knock sensor signal is on the positive side of the vibration center level, and is from the time when the affirmative determination in S230 to the affirmative determination in S220. The period can be regarded as a negative half-wave period in which the level of the knock sensor signal is more negative than the vibration center level. And further, 2 μs in FIG.
In each process, in the processes of S290 and S240, from the time when it is determined that the level of the knock sensor signal has crossed the reference value by the processes of S220 and S230 as a starting point, until the next determination is made that the level has crossed the reference value. ,
The absolute value of "NEWAD" is cumulatively added. Then, by the processing of S290 and S240,
The swing width of the knock sensor signal with respect to the reference value is integrated over a half-wave period. For this reason, in the case where an affirmative determination is made in either of S220 and S230, the value stored in the integrated value storage area SUM is the integrated value obtained by integrating the swing width with respect to the reference value of the knock sensor signal by a half-wave period. This integral value is calculated for each half-wave period of the knock sensor signal.

In the processing of S300 and S250,
The number of times the absolute value of “NEWAD” is added in the processing of S290 and S240, and the duration of the half-wave period corresponding to the integration time is counted as the value of the counter N. Then, in the process of S260 performed each time an affirmative determination is made in any of S220 and S230,
The value stored in the integrated value storage area SUM (that is, the integrated value of the swing width for the half-wave period with respect to the reference value of the knock sensor signal) is used as the value of the counter N as the duration of the half-wave period as the integration time The average value AVR of the swing width for the half-wave period with respect to the reference value of the knock sensor signal is calculated.

Therefore, every time the process of S260 is performed,
As shown in FIG. 7, the average value AVR of the swing width with respect to the reference value of the knock sensor signal for each half-wave period T is sequentially calculated. Further, in the processing of S270 and S280, the average value AVR calculated this time in S260 is
If the average value AVRmax is larger than the previous maximum value "AVRmax" (S270: YES), the current average value AVR is updated and stored as the maximum average value "AVRmax" (S280).

Therefore, "AVRmax" at the time when the processing is completed every 2 .mu.s becomes the maximum value among the average values AVR calculated in S260, as shown in FIG. Next, FIG. 5 is a flowchart showing a knock determination section stop process executed by the second CPU 29 at the timing of ATDC 90 ° CA.

As shown in FIG. 5, when the second CPU 29 starts execution of the knock determination section stop processing, first, at S4
At 00, the first CPU 27 is prohibited from executing the process every 2 μs. Then, thereafter, the first CPU 27
Until the PU 29 next executes S110 of the knock determination section start process, the process is not executed every 2 μs in FIG.

Then, the second CPU 29 proceeds to S410
In the processing performed by the first CPU 27 every 2 μs,
The "AVRmax" finally stored in the "AVRmax" is read. At 420, both the positive direction and the negative direction with respect to the reference value of the knock sensor signal are calculated from the read "AVRmax" using the following equation 5. The maximum peak value P / Hmax, which is the maximum value of the swing width to the maximum value, is calculated.

[0069]

P / Hmax = (π / 2) × AVRmax Equation 5 Equation 5 represents the same meaning as Equation 2 described above.
Among the average values AVR calculated in S260 of FIG.
This shows that the peak value Vp in the half-wave period corresponding to the maximum average value AVRmax is the maximum peak value P / Hmax.

Further, in subsequent S430, the above S4
It is determined whether the maximum peak value P / Hmax calculated in step 20 is greater than a predetermined knock determination value (Ko × Av). If the maximum peak value P / Hmax is greater than the knock determination value (Ko × Av), Proceeding to S440, a knock determination flag indicating the occurrence of knock is set, and conversely, the maximum peak value P /
If Hmax is not larger than the knock determination value (Ko × Av), the flow shifts to S450, where the knock determination flag is reset.

When the knock determination flag is set in the process of S440, the second CPU 29 suppresses the occurrence of knock by correcting the ignition timing of the engine to the retard side as described above. Further, among the variables constituting the knock determination value (Ko × Av), Ko is an arbitrary constant, and Av is the average value of the maximum peak values P / Hmax up to the previous knock determination section.

Then, after performing one of the processes in S440 and S450, the flow shifts to S460, and the average value Av is updated as in the following Expression 6. That is, in the present embodiment, the average value Av is updated based on the so-called 1/8 smoothing concept.

[0073]

Av = (7 × Av + P / Hmax) / 8 Formula 6 After performing the process of S460, the knock determination section stop process ends.

As described in detail above, E of the first embodiment is
The CU 1 integrates the absolute value of “NEWAD” representing the amplitude of the knock sensor signal with respect to the reference value for each half-wave period of the knock sensor signal (S220, S23).
0, S290, S240), and dividing each integral value by the value of the counter N representing the duration of the half-wave period, thereby obtaining the average value AVR of the swing width with respect to the reference value of the knock sensor signal for each half-wave period. Is calculated (S260). The largest one of the average values AVR is referred to as “AVR”.
max ”(S270, S280), and“ A ”
VRmax "and the maximum peak value P / Hmax using Equation 5.
Is calculated (S420).

In the ECU 1 according to the first embodiment, the swing width of the knock sensor signal with respect to the reference value is integrated and averaged for each half-wave period. Therefore, as shown in the upper part of FIG. Even if noise is superimposed on a signal, the influence of the noise can be kept very small.
That is, even if the peak value of the noise is large, the ratio of the noise component in the integrated value is small, and the influence of the noise does not significantly appear in the average value AVR.

Further, in the ECU 1 according to the first embodiment, the swing width of the knock sensor signal with respect to the reference value is not integrated over the knock determination section, but is integrated every half-wave period of the knock sensor signal. Therefore, regardless of the length of the knock determination section, the maximum peak value P / Hmax of the knock sensor signal in the knock determination section can be accurately detected.

Thus, according to the ECU 1 of the first embodiment, the maximum peak value P / Hmax of the knock sensor signal can be accurately detected without being affected by noise and regardless of the length of the knock determination section. . Therefore, it is possible to accurately determine whether or not the engine is knocked.

Further, in the ECU 1 of the first embodiment, the number of times the absolute value of "NEWAD" is added, and the duration of the half-wave period as the integration time (more specifically, the duration is divided by 2 μs). Is measured by the counter N (S300, S250), and the average value AVR is calculated by dividing the integrated value of the swing width with respect to the reference value of the knock sensor signal for the half-wave period by the value of the counter N. (S260).

Therefore, even if the cycle of the knock sensor signal (especially, the actual duration of the half-wave period) changes, S26 in FIG.
Each average value AVR calculated with 0 is always an accurate value,
Therefore, the maximum peak value P / Hmax of the knock sensor signal
Can be detected more accurately. That is, the divisor used for obtaining the average value AVR in S260 of FIG. 4 is a fixed value corresponding to a normal half cycle time of the knock sensor signal (for example, a value obtained by dividing a half cycle of 8 kHz by 2 μs.
Although 1) may be used, as in the present embodiment, if the actual duration of the half-wave period is measured and the measured value is used as a divisor, the average value AVR can be obtained more accurately. Is advantageous.

In the first embodiment, S220 in FIG.
4 and S230 correspond to a half-wave period specifying unit, and the processes of S290 and S240 in FIG. 4 correspond to an integrating unit. Then, the process of S260 in FIG. 4 corresponds to the average value calculating means, and the processes of S270 and S280 in FIG.
The processes of S410 and S420 in FIG. 5 correspond to the maximum value calculating means. Further, S300 and S300 in FIG.
Step 250 corresponds to the time detecting means.

[Second Embodiment] The first embodiment described above
In the ECU 1 of the embodiment, the half-wave period of the knock sensor signal is specified as a period from when the level of the knock sensor signal crosses the reference value to when it crosses the reference value again,
The duration of the half-wave period specified in this way is determined by a counter N
Is measured by the value of.

Therefore, as shown from time t1 to time t4 in FIG. 9, if the level of the knock sensor signal crosses the reference value due to noise despite the fact that it is actually in a half-wave period, S220 and S23 in FIG.
The duration of the half-wave period identified by 0 becomes shorter,
Accordingly, the value of the counter N used as the divisor in S260 of FIG. 4 becomes small.

Then, there is a possibility that the average value AVR calculated in S260 of FIG. 4 becomes excessively large.
There is a possibility that the maximum peak value P / Hmax calculated at 20 becomes larger than the true maximum peak value. Then, in order to solve such a problem, E of the second embodiment is used.
In the CU, the first CPU 27 executes the processing every 2 μs in FIG. 10 instead of the processing every 2 μs in FIG. Then, the processing every 2 μs in FIG. 10 is different from the processing every 2 μs in FIG. 4 only in that the processing of S500 is added. Note that, in FIG. 10, the same processes as those in FIG. 4 are denoted by the same step numbers, and a detailed description thereof will be omitted.

That is, as shown in FIG. 10, in the process every 2 μs of the second embodiment, if the judgment is affirmative in either of S220 and S230 (S220: YES or S2
30: YES), proceed to S500, and the value of the counter N at that time is within a set range (N1 to N2) from a predetermined first set value N1 to a second set value N2 (> N1). It is determined whether or not. In the present embodiment, the first
The set value N1 is a half cycle of 10 kHz (= 50 μs) = 2
It is set to 24 slightly smaller than the value divided by μs, and the second set value N2 is a half cycle of 6 kHz (= 83.3).
μs) is set to 42 which is slightly larger than a value obtained by dividing 2 μs by 2 μs.

If it is determined in step S500 that the value of the counter N is within the set range (N1 to N2), the process proceeds to step S260. In contrast, S
At 500, the value of the counter N is set in the setting range (N1 to N
If it is determined that it is not within 2), the process directly proceeds to S290. Then, the integral value stored in the integral value storage area SUM at the time of execution of S500 is discarded, and the calculation of the average value AVR for the integral value is not performed. As a result, the integral value becomes the maximum peak value. P / Hmax
Will be excluded from the target for calculating.

Therefore, for example, in FIG. 9, a time interval t2 to t3 at which the level of the knock sensor signal crosses the reference value due to noise, and a time interval t3 to t4 at which the original half-wave period is recovered from the noise and ended. And 10 kh
When the period becomes shorter than the half cycle of z, the average value AVR is not calculated for the integrated value at each of the time intervals t2 to t3 and t3 to t4, and further, the calculated maximum peak value It is possible to reliably prevent P / Hmax from becoming larger than the true value.

Therefore, the ECU according to the second embodiment as described above
According to this, the influence of noise is further reduced, and the maximum peak value P / Hmax of the knock sensor signal can be detected more accurately. In addition, E of the second embodiment
According to the CU, for example, from time t4 to time t in FIG.
As shown in FIG. 5, even if the half-wave period of the knock sensor signal becomes longer than the half period of 6 kHz for some reason, the average value of the integrated value from time t4 to time t5 is not changed. It is possible to prevent the value AVR from being calculated and adversely affecting the calculated maximum peak value P / Hmax.

In the second embodiment, S50 in FIG.
The process of 0 corresponds to the determination means. [Third Embodiment] Next, an ECU according to a third embodiment will be described. In the third embodiment described below, “OLDA” in the first and second embodiments described above.
"D" is described as "OLDAD1". And "O
LDAD2 ”is a detection value that is one time earlier than“ OLDAD1 ”(that is, the level of the knock sensor signal is set to A
(Detection value obtained by subtracting the reference value from the A / D converted value after the A / D conversion)
It is.

First, the ECU of the third embodiment differs from the ECU 1 of the first embodiment in the following (A) and (A).
(B) is different. (A) First, the second CPU 29 determines “OLD” in the RAM 33 in S100 of the knock determination section start process in FIG.
The value of "AD1" is also initialized to zero.

(B) Then, the first CPU 27 executes the processing every 2 μs shown in FIG. 11 instead of the processing every 2 μs in FIG. Therefore, next, the ECU of the third embodiment performs the first C
The processing performed by the PU 27 every 2 μs will be described with reference to FIG.

As shown in FIG. 11, E of the third embodiment is
In the CU, when the first CPU 27 starts executing the process every 2 μs, first in S600, the “OLDA” up to that point is reached.
D1 ”is stored again as“ OLDAD2 ”, and the subsequent S6
At 10, the “NEWAD” was replaced by “OLDAD”
Re-store as "1". Further, in S620, a command is given to the multiplexer 23 and the A / D converter 2
5, a knock sensor signal is input, and the level of the knock sensor signal is A / D-converted by an A / D converter 25, and a reference value (= 2.5 V) is obtained from the current A / D converted value.
Is stored as “NEWAD”.

That is, in the third embodiment, the value obtained by subtracting the reference value from the A / D converted value of the level of the knock sensor signal is 2 μm.
Among the detected values for each s, three consecutive values (“NEWAD”, “OLDAD1”, “OLDAD1”)
2 "). Next, the following S630
It is determined whether or not the relationship of the following equation 7 holds. That is, in S630, as shown in FIG. 12A, three detection values (“OLDAD2”, “OLDAD1”, and “NEDAD1”) continuously detected by the process of S620.
WAD ”), the second detection value (“ OLDAD
1 ") is determined to be 0 or less.

[0093]

(OLDAD2)> 0 ≧ “OLDAD1” and “NEWAD”> 0 Expression 7 If it is determined in S630 that the relationship of Expression 7 is not established, the process proceeds to S640. This time, it is determined whether or not the relationship of the following equation 8 holds. That is, in this S640, as shown in FIG.
The three detection values (“OL
It is determined whether or not only the second detection value (“OLDAD1”) among DAD2 ”,“ OLDAD1 ”, and“ NEWAD ”is 0 or more.

[0094]

"OLDAD2"<0≤"OLDAD1" and "NEWAD"<0 ... Equation 8 Here, if it is determined in S640 that the relationship of Equation 8 is not established, the flow proceeds to S650. , S220 in FIG.
Similarly to the above, it is determined whether or not the relationship of the following equation 9 holds, thereby determining whether or not the level of the knock sensor signal has crossed the reference value from the negative side to the positive side.

[0095]

"OLDAD1" ≤0 <"NEWAD" (Equation 9) If it is determined in S650 that the relationship of Equation 9 is not established, the process proceeds to S660 and proceeds to S230 of FIG.
Similarly to the above, it is determined whether or not the relationship of the following Expression 10 is established, thereby determining whether or not the level of the knock sensor signal has crossed the reference value from the positive side to the negative side.

[0096]

"OLDAD1" ≧ 0> “NEWAD” (Equation 10) When it is determined in S660 that the relationship of Expression 10 is not established, that is, when the level of the knock sensor signal does not cross the reference value In S670, the process proceeds to S670, and the value stored in the integral value storage area SUM is displayed as “OLDAD”.
1 ”(ie, the amplitude of the knock sensor signal detected last time relative to the reference value), and the value after the addition (=
| OLDAD1 | + SUM) is stored again in the integral value storage area SUM. Then, in S680, the value of the counter N for counting the number of additions is incremented by one, and then the process is terminated once every 2 μs.

When it is determined in S650 that the relationship of Expression 9 is established, or in S660, Expression 10 is satisfied.
When it is determined that the relationship is established, it is determined that the level of the knock sensor signal has crossed the reference value, and S69
Move to 0. Then, in S690, similarly to S670, the absolute value of “OLDAD1” is added to the value stored in the integration value storage area SUM, and the value after the addition (= | OLDAD1 | + SUM) is stored in the integration value storage. Area SU
Re-store it in M. Further, in subsequent S700, the value of the counter N is incremented by one.

Then, in subsequent S710, S26 in FIG.
Similarly to 0, by dividing the value stored in the integrated value storage area SUM at that time by the value of the counter N, the average value AVR (= SUM) of the amplitude for the half-wave period with respect to the reference value of the knock sensor signal is obtained. / N).

Next, similarly to S270 and S280 in FIG. 4, in S720, it is determined whether the average value AVR calculated in S710 is larger than "AVRmax".
If the average value AVR is not larger than "AVRmax" (S
720: NO), the process directly proceeds to S740, but the average value A
If VR is larger than “AVRmax” (S720: YE
S), in the next S730, the average value AVR is set to “AVR
max ”and then proceed to S740.

Then, in S740, the integral value storage area S
0 is stored in UM as an initial value, and in S750, the value of the counter N is initialized to 0. Then, thereafter, the process is terminated once every 2 μs. On the other hand, if it is determined in S630 that the relationship of Expression 7 is established, or if it is determined in S640 that the relationship of Expression 8 is established, the process proceeds to S760, and “OLDAD2” is set. "NEW
AD ”(= (OLDAD2 + NEWAD) /
2) is stored again as “OLDAD1”, and then
The process moves to S670. For this reason, in S670 in this case, instead of the original absolute value of “OLDAD1”, “OLD
The absolute value of the average value of “AD2” and “NEWAD” is cumulatively added to the value of the integrated value storage area SUM.

That is, in the process every 2 μs of the third embodiment, it is determined whether or not the level of the knock sensor signal has crossed the reference value by the processes of S650 and S660, and the level of the knock sensor signal is set to the reference value. A period from the time when it is determined that the vehicle has crossed the vehicle to the time when it is determined that the vehicle has crossed the reference value again. Is determined as a half-wave period of the knock sensor signal.

Further, S740, S670, and S
By the processing of 690, the absolute value of “OLDAD1” is determined from the time when it is determined that the level of the knock sensor signal has crossed the reference value by the processing of S650 and S660 until the next determination is made that the signal has crossed the reference value. The values are cumulatively added. And this S740, S
By the processes of 670 and S690, the amplitude of the knock sensor signal relative to the reference value is integrated over a half-wave period. For this reason, immediately after the processing of S690 is performed after an affirmative determination in either of S650 and S660, the value stored in the integrated value storage area SUM is reduced by half the amplitude of the knock sensor signal with respect to the reference value. The integrated value is obtained by integrating for the wave period, and this integrated value is calculated for each half-wave period of the knock sensor signal.

Also, in the processing every 2 μs of the third embodiment, the processing of S750, S680, and S700 is the number of additions of the absolute value of “OLDAD1” in the processing of S740, S670, and S690. The duration of the half-wave period corresponding to the time is counted as the value of the counter N.

Each time a positive determination is made in either of S650 and S660, the integrated value storage area S is executed by the processing of S710 performed immediately after S690 and S700.
The value stored in the UM (that is, the integral value of the swing width with respect to the reference value of the knock sensor signal for the half-wave period) is divided by the value of the counter N as the duration of the half-wave period, which is the integration time, The average value AVR of the swing width for the half-wave period with respect to the reference value of the knock sensor signal is calculated. Therefore, also in the third embodiment, every time the process of S710 is performed, as shown in FIG. 7, the average value AVR of the swing width with respect to the reference value of the knock sensor signal for each half-wave period T is sequentially calculated. It will go.

Further, in the processing of S720 and S730, the average value AVR calculated this time in S710 is compared with the maximum average value “AVRmax” up to the previous time, and if the average value AVR calculated this time is larger, (S720: Y
ES), and the current average value AVR is set to the maximum average value “AV
Rmax "is updated and stored (S73).
0).

Therefore, also in the third embodiment, FIG.
“AVRmax” at the time when the processing of 1 is completed every 2 μs is
As shown in FIG. 7, each average value AV calculated in S710
R is the largest one. Then, the second CPU 29
However, in S420 in FIG. 5, the maximum peak value P / Hmax is calculated using “Equation 5” from “AVRmax” at the time when the processing at every 2 μs in FIG. 11 is completed.

In particular, in the processing every 2 μs of the third embodiment, three consecutive detection values (“OLDAD2”, “OLDAD1”, “NEWA”) are obtained by the processing of S630.
D ”), the second detection value (“ OLDAD1 ”)
It is determined whether or not only the second detection value (“OLDAD1”) among the three consecutive detection values is 0 or more by the process of S640. Has been determined.

If an affirmative determination is made in either of S630 and S640, the process of S760 is performed, and
As shown by the squares in FIG. 12, the above three detection values (“OLDAD2”, “OLDAD1”, “NEWA”)
D ”) of the first detection value (“ OLDAD2 ”) and the third detection value (“ NEWAD ”) (= (O
LDAD2 + NEWAD) / 2) is replaced with the second detection value (“OLDAD1”), whereby the absolute value of the average value (= (OLDAD2 + NEWAD) / 2) is replaced with the original absolute value of “OLDAD1”. , Are cumulatively added to the value of the integral value storage area SUM.

Therefore, according to the ECU of the third embodiment, not only effects similar to those of the ECU 1 of the first embodiment can be obtained, but also noise is superimposed on the sensor signal, and the ECU shown in FIG.
As shown in (a), three consecutive detection values (“OLDAD
2 ”,“ OLDAD1 ”, and“ NEWAD ”), only the second detection value (“ OLDAD1 ”) becomes 0 or less, or as shown in FIG. 12B, the second detection value (“ OLAD1 ”).
Even if only DAD1 ") becomes 0 or more, the influence of such noise can be avoided, and the maximum peak value P / Hmax of the knock sensor signal can be accurately detected.

In the third embodiment, S62 in FIG.
0 corresponds to the level detecting means, and corresponds to S65 in FIG.
0 and S660 correspond to the half-wave period specifying means,
The processing of S740, S670, and S690 of FIG.
This corresponds to the integration means. Then, S750 of FIG.
The processing of S680 and S700 corresponds to the time detecting means, the processing of S710 in FIG. 11 corresponds to the average value calculating means, and the processing of S720 and S730 in FIG.
Steps 410 and S420 correspond to the maximum value calculating means. Further, S630, S640,
The processing of S760 and S760 corresponds to a correction unit.

As described above, one embodiment of the present invention has been described. However, it is needless to say that the present invention is not limited to the above-described embodiment, and can take various forms. (1) For example, in each of the above embodiments, each average value AV of the swing width with respect to the reference value of the knock sensor signal for each half-wave period
The largest one among R is selected as “AVRmax”, and the maximum peak value P / P is determined from the selected “AVRmax”.
Hmax was calculated. For each average value AVR, the peak value Vp during the half-wave period was calculated using Expression 2 (specifically, replacing Va in Expression 2 with AVR), and each of the calculated values was calculated. The maximum peak value Vp may be selected as the maximum peak value P / Hmax. However, if the procedure of each of the above embodiments is adopted, the average value AVR
Calculation processing for calculating the peak value in the half-wave period from
This is advantageous because it only needs to be performed once.

(2) In the above embodiments, the maximum value of the swing width in both the positive direction and the negative direction with respect to the reference value of the knock sensor signal is detected as the maximum peak value P / Hmax. . On the other hand, for example, in the first embodiment, when the maximum value of the swing width in the positive direction with respect to the reference value of the knock sensor signal is detected as the maximum peak value P / Hmax, the following change is made. good.

First, in the processing every 2 μs in FIG.
The flag FP indicating the positive half-wave period is set immediately after the affirmative determination at 20, and the flag FP is reset immediately after the affirmative determination at S230. Then, the processing from S260 to S280 is performed when the flag FP is reset (that is, when the period changes from the positive half-wave period to the negative half-wave period), and the processing of S290 and S300 is performed. And the processing of S240 and S250 are performed only when the flag FP is set (that is, only during the positive half-wave period).

By changing in this way, S290 and S2
The integration by 40 (that is, the cumulative addition of the absolute value of "NEWAD") and the measurement of the integration time by S300 and S250 are performed only during the positive half-wave period, and from the positive half-wave period to the negative half-wave period. When switching to the wave period, the average value AVR is calculated from the integrated value for the positive half-wave period, and the maximum value of the swing in the positive direction with respect to the reference value of the knock sensor signal is set to the maximum value. It can be detected as the peak value P / Hmax.

In the second embodiment, the processes from S500 and S260 to S280 in FIG. 10 may be performed when the flag FP is reset.
Similarly, in the third embodiment, when the maximum value of the swing width in the positive direction with respect to the reference value of the knock sensor signal is detected as the maximum peak value P / Hmax, the following change may be made.

First, in the processing every 2 μs in FIG.
The flag FP indicating the positive half-wave period is set immediately after the affirmative determination in 650, and the flag FP is reset immediately after the affirmative determination in S660. Then, the processing from S690 to S730 is performed when the flag FP is reset (that is, when the period changes from the positive half-wave period to the negative half-wave period), and the processing of S740 and S750 is performed. And the processes of S670 and S680 may be performed only when the flag FP is set (that is, only during the positive half-wave period).

(3) Conversely, for example, in the first embodiment, when the maximum value of the swing width in the negative direction with respect to the reference value of the knock sensor signal is detected as the maximum peak value P / Hmax, It can be changed as follows. First, FIG.
In the process of 2 μs, the flag FM indicating the negative half-wave period is set immediately after the affirmative determination in S230, and the flag FM is reset immediately after the affirmative determination in S220.

The processing from S260 to S280 is performed when the flag FM is reset (that is, when the period changes from the negative half-wave period to the positive half-wave period). Processing of S300 and S2
Steps S40 and S250 are performed only when the flag FM is set (that is, only during the negative half-wave period).

By changing in this way, S290 and S2
The integration by 40 and the measurement of the integration time by S300 and S250 are performed only during the negative half-wave period,
When the transition is made from the negative half-wave period to the positive half-wave period, the average value AVR is calculated from the integral value for the negative half-wave period, and hence in the negative direction with respect to the reference value of the knock sensor signal. Can be detected as the maximum peak value P / Hmax.

In the second embodiment, the processes from S500 and S260 to S280 in FIG. 10 may be performed when the flag FM is reset.
Similarly, in the third embodiment, when the maximum value of the swing width in the negative direction with respect to the reference value of the knock sensor signal is detected as the maximum peak value P / Hmax, the following change may be made.

First, in the process every 2 μs in FIG.
Immediately after the affirmative determination in 660, the flag FM indicating the negative half-wave period is set, and immediately after the affirmative determination in S650, the flag FM is reset. Then, the processing from S690 to S730 is performed when the flag FM is reset (that is, when the period changes from the negative half-wave period to the positive half-wave period), and the processing of S740 and S750 is performed. And the processes of S670 and S680 may be performed only when the flag FM is set (that is, only during the negative half-wave period).

(4) In each of the above embodiments, the amplitude of the knock sensor signal relative to the reference value may be continuously integrated by an analog integrating circuit. (5) Further, the ECU of each of the above embodiments determines whether or not there is knock by comparing the maximum peak value P / Hmax with the knock determination value (Ko × Av) in S430 of FIG. However, in S430 of FIG. 5, S4
The magnitude of the "AVRmax" read in step 10 may be compared with a predetermined determination value, and if "AVRmax" is greater, it may be determined that knock has occurred and a knock determination flag may be set. Also in this case, the knock of the engine can be accurately detected without being affected by noise and regardless of the length of the knock determination section.

(6) On the other hand, in each of the above embodiments, the case where only one knock sensor 3 is attached to the engine has been described. However, when a plurality of knock sensors 3 are attached like a V-type engine, The above-described processing may be performed for each knock sensor.

(7) In each of the above-described embodiments, the knock sensor 3 is a vibration-type sensor mounted on the cylinder block of the engine. The same can be applied to a knock sensor that outputs a corresponding vibration wave.

(8) Furthermore, the ECU of each of the above embodiments
Is intended to detect a sensor signal from the knock sensor 3, but the present invention can be similarly applied to any device that detects a sensor signal as a vibration wave.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram illustrating an example of an operation of the present invention.

FIG. 2 is a block diagram illustrating a configuration of an electronic control unit (ECU) according to the first embodiment.

FIG. 3 is a flowchart illustrating a knock determination section start process executed by a second CPU.

FIG. 4 is a flowchart showing a process performed by a first CPU every 2 μs.

FIG. 5 is a flowchart illustrating a knock determination section stop process executed by a second CPU.

FIG. 6 is an explanatory diagram illustrating a knock sensor signal output from a knock sensor and a knock determination section.

FIG. 7 is an explanatory diagram for explaining the operation of the process every 2 μs in FIG. 4;

FIG. 8 is an explanatory diagram showing a part of FIG. 7 in an enlarged manner.

FIG. 9 is an explanatory diagram illustrating a problem to be solved by an ECU according to a second embodiment.

FIG. 10 shows a second embodiment of the second embodiment executed by the first CPU.
It is a flowchart showing a process for every microsecond.

FIG. 11 shows a second embodiment of the third embodiment executed by the first CPU.
It is a flowchart showing a process for every microsecond.

FIG. 12 is an explanatory diagram for explaining the operation of the process performed every 2 μs in FIG. 11;

FIG. 13 is an explanatory diagram for explaining a problem of the related art.

[Explanation of symbols]

1. Electronic control unit (ECU) 3. Knock sensor
5 Band-pass filter 7 Rotation angle sensor 11 Input circuit 23 Multiplexer 25 A / D converter 27 First CPU 29 Second CPU 31 ROM 33 RAM 35 Microcomputer

Claims (8)

[Claims] 1. A sensor signal output as a vibration wave from a sensor is input for a predetermined detection period, and the sensor signal swings in a positive direction and / or a negative direction with respect to a vibration center level during the detection period. A method of detecting a maximum value of a sensor signal for detecting a maximum value of a width, wherein a swing width of the sensor signal with respect to the vibration center level is a half-wave period in which the level of the sensor signal is on the positive side of the vibration center level And the level of the sensor signal is on the negative side of the oscillation center level for each of both of the half-wave periods, or for each of the two half-wave periods. Is calculated by dividing the average value by the duration of the half-wave period, and the maximum value of the amplitude is obtained from the maximum value among the calculated average values. Method. 2. The method according to claim 1, wherein the sensor is a knock sensor for detecting a knock generated in an internal combustion engine. Method. 3. A sensor signal output as a vibration wave from a sensor is input during a predetermined detection period, and the sensor signal swings in a positive direction and / or a negative direction with respect to a vibration center level during the detection period. A sensor signal processing device that detects a maximum value of a width, wherein a half-wave period in which the level of the sensor signal is more positive than the vibration center level and the level of the sensor signal is more negative than the vibration center level. A half-wave period specifying unit that specifies both or one of the half-wave periods; and a swing width of the sensor signal with respect to the oscillation center level is integrated for each half-wave period specified by the half-wave period specifying unit. Integrating means; average value calculating means for calculating an average value obtained by dividing the integrated value by the integrating means by the duration of the half-wave period; and each average calculated by the average value calculating means. Maximum one from the shake sensor signal processing apparatus characterized by comprising: a maximum value calculating means, the calculating the maximum value of the width of the. 4. The sensor signal processing device according to claim 3, further comprising: a time detecting unit configured to detect a duration of the half-wave period specified by the half-wave period specifying unit; The average value is calculated by dividing the integrated value by the integrating means by the duration detected by the time detecting means. 5. The sensor signal processing device according to claim 4, wherein said half-wave period specifying means sets said half-wave period to said vibration center level again after a level of said sensor signal crosses said vibration center level. It is configured to specify as a period until crossing, further, the device determines whether the duration detected by the time detecting means is within a predetermined setting range, If the duration is not within the set range, the sensor signal includes a determination unit that discards an integrated value of the integration unit corresponding to a half-wave period of the duration that is not within the set range. Processing equipment. 6. The sensor signal processing device according to claim 3, further comprising a level detection unit that detects a difference between the level of the sensor signal and the vibration center level at regular time intervals, The integration means is configured to integrate the amplitude of the sensor signal with respect to the vibration center level by sequentially adding the absolute values of the detection values obtained by the level detection means. When only the second detection value is 0 or less among the three detection values successively detected by the level detection means, or when only the second detection value is 0 or more, the second Correcting means for adding the absolute value of the average of the first and third detection values of the three detection values to the integration means, instead of the absolute value of the third detection value; Features Sensor signal processing apparatus. 7. The sensor signal processing device according to claim 3, wherein the sensor is a knock sensor for detecting a knock generated in the internal combustion engine. Signal processing device. 8. A sensor signal output as a vibration wave from a knock sensor for detecting knock generated in the internal combustion engine is inputted for a predetermined detection period, and the knock of the internal combustion engine is obtained from the sensor signal in the detection period. Wherein the amplitude of the sensor signal with respect to the vibration center level is a half-wave period in which the level of the sensor signal is more positive than the vibration center level, and the level of the sensor signal is the vibration. Integrate every minute for both of the half-wave periods on the negative side of the center level, or for each one of the two half-wave periods, and integrate each integral value with the duration of the half-wave period. A knock detection method for an internal combustion engine, comprising: calculating a divided average value; and detecting a knock of the internal combustion engine from a maximum value among the calculated average values.