JP3117495B2 - Knock detection device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a knocking detection device for an internal combustion engine, and more particularly to a knocking detection device suitable for an automobile gasoline engine.

[0002]

2. Description of the Related Art Knocking on an engine (also known as knocking)
Occurs, a vibration having a specific resonance frequency component is generated. Therefore, the occurrence of knocking can be determined by detecting this vibration with a knock sensor.
Therefore, in the conventional knocking detection device, a band-pass filter is used to extract only a specific frequency component having the highest frequency among the frequency components included in the signal from the knock sensor. Was determined by the size of the knocking. An apparatus related to this is disclosed in Japanese Patent Application Laid-Open No. 59-737.
No. 50, JP-A-59-125034, JP-A-60-2
No. 04969 and JP-A-1-178773.

[0003]

In the above prior art, no consideration is given to the fact that knocking has a plurality of vibration modes, and some of the vibration modes when knocking occurs are processed as noise. However, there is a limit in the knocking detection ability, the operating range of the engine in which knocking can be detected is limited, and there is a problem in reliable knocking detection.

That is, in the conventional apparatus, only a specific frequency is detected from signals included in a knock sensor signal using a band-pass filter. However, in an actual knocking phenomenon, only a single frequency component is detected. Does not occur, but a plurality of knocking vibration modes exist.

That is, as shown in FIG. 6, the cylinder of the engine is cylindrical as its name suggests. Therefore, as vibration modes in the cylinder, nodes (partition lines) of pressure fluctuations and antinodes (+, −), And as a result, five types of frequency components appear as a resonance phenomenon in the cylinder due to knocking. However, according to the conventional technique, only a specific frequency that is most likely to resonate is extracted from these frequencies, and therefore, the knocking detection capability is limited.

[0006] The frequency of knocking vibration also changes depending on the properties of gas in the combustion chamber. In other words, the resonance wavelength of the cylinder is determined by the shape of the combustion chamber that forms the resonance cavity. However, if the combustion temperature or the compression ratio increases, the speed of sound in the combustion gas increases, and the frequency increases. Further, when the compression ratio of the engine increases, not only does the combustion temperature rise, but also the shape of the resonance cavity changes, and the resonance mode also changes.

In this respect, the technique using the conventional band-pass filter can cope with the problem by widening the band of the filter. As a result, noise can be easily picked up, and the knocking can be reliably detected. There was a problem in the point. It is an object of the present invention to provide a knocking detection device that can always reliably determine knocking irrespective of the operating state of an engine and can easily perform accurate knocking control.

[0008]

An object of the present invention is to provide an engine
Detects at least one of vibration and cylinder pressure vibration
At least two predetermined signals included in the output signal of the knock sensor
Frequency component extracting means for independently extracting the frequency components of
And at least the frequency component extraction means
Knock finger that calculates the knock index from the operation result of the output of 2.
Number arithmetic means, and converts the at least two frequency components
Each resonance mode that appears in the engine cylinders
By matching the resonance frequency component determined by
Knocking method that determines knocking occurrence from knock index
In the at least two frequency components.
Extract at least three signals of similar frequency for each
Frequency analysis means including a digital filter
Having a maximum amplitude level of at least three signals
The center frequency of the frequency analysis means according to the frequency of the signal
Depending on the operating condition of the engine
Correction means for changing the frequency of the digital filter
And the center frequency of the frequency analysis
It is automatically tracked by the frequency change of the frequency component,
The frequency analysis section by the filter is the operating state of the engine
Is achieved in such a way as to be varied in accordance with

[0009]

The vibration detecting means can detect all modes of knocking vibration generated in the combustion chamber, and the frequency component extracting means can separately detect each vibration mode. At this time, by performing a knock determination for each vibration mode, a signal that has been conventionally excluded as a noise component can also be taken into knock detection, so that knock detection capability can be improved. And the center frequency of the frequency component extraction means automatically tracks according to the change of the vibration mode,
Knock can always be reliably determined, and accurate knock control can be performed.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a knocking detection device according to the present invention will be described in detail with reference to the illustrated embodiment. FIG. 5 is a configuration diagram of an engine control device to which an embodiment of the present invention is applied. In the drawing, reference numeral 1 denotes an engine control unit, 2 denotes an engine cylinder, 3 denotes an intake pipe, 4 denotes an exhaust pipe, and 5 denotes a throttle. A valve, 6 is a fuel injection valve, 7 is a spark plug, 8 is an ignition coil, 9 is a distributor, 10 is an air flow sensor for measuring an intake air flow rate, 11 is a reference sensor for distinguishing a cylinder, and 12 is a rotation angle. Position sensor, 13
Is an opening sensor for detecting a throttle valve, 14 is an O ₂ sensor for detecting an air-fuel ratio, 15A is a knock vibration sensor, 15B is a seat pressure sensor, and 15C is a cylinder pressure sensor.

The engine control unit 1 is composed of various electronic circuits including a microcomputer, and includes an intake air flow meter 10, a reference angle sensor 11, a position angle sensor 12, and a throttle valve opening sensor 1.
3. It takes in signals from various sensors required to know the operating state of the engine, such as the air-fuel ratio sensor 14,
Various drive signals are generated by predetermined arithmetic processing, and the fuel injection valve 6 and the ignition coil 8 are operated to control the engine.

Knock vibration sensor 15A and seat pressure sensor 15
B and the in-cylinder pressure sensor 15C are all provided as knock signal detecting means. First, the knock vibration sensor 15A is attached near the combustion chamber of the cylinder block of the engine, and detects the vibration accompanying the combustion of the engine. Next, the seat pressure sensor 15B is provided in the washer portion of the ignition plug 7 and functions to directly capture vibration in the combustion chamber and convert the vibration into an electric signal. And
The in-cylinder pressure sensor 15C is provided with a hole in a part of the cylinder head of the combustion chamber, and also functions to directly capture the vibration in the combustion chamber and convert it into an electric signal.

In the embodiment of the present invention, only one knock signal detecting means is required. Therefore, in practice, any one of knock vibration sensor 15A, seat pressure sensor 15B and in-cylinder pressure sensor 15C is provided. Only one will be used.

By the way, as described above, the vibration mode and the frequency at the time of occurrence of knocking are as shown in FIG. That is, as the vibration of the cylinder in the combustion chamber, a resonance state occurs at the nodes (separation lines) and antinodes (+ or-) of the vibration in the radial direction and the circumferential direction, and only a specific frequency component is generated.
Since the knocking signal detecting means detects the vibration generated by the knocking generated in this way, its output includes signals of a plurality of vibration modes.

Therefore, it is desirable that the knock vibration sensor 15A, the washer sensor 15B, or the in-cylinder pressure sensor 15C serving as the knock signal detecting means have uniform sensitivity over the entire band including the frequency shown in FIG. In addition, a piezo-type piezoelectric detection sensor using piezoelectric ceramic or crystal is generally used, and therefore, such a sensor is also used in this embodiment.

As described above, the engine control unit 1
Supplies a drive signal to the fuel injection valve 6 and the ignition coil 8 to control the engine. In this embodiment, the knocking vibration sensor 15A, the seat pressure sensor 15B, and the cylinder pressure sensor 15C A knocking determination process is performed by taking in a signal (knock signal) from the knocking signal detecting means consisting of one or more, and based on the result, the knocking control given by delaying the ignition timing in a predetermined manner is performed. The knock determination process performed by the engine control unit 1 at this time will be sequentially described below.

First, a signal from the knock signal detecting means (hereinafter referred to as knock sensor 15) is input to a predetermined frequency analyzing means, and is separated and extracted into respective frequency components.

By the way, as the frequency analysis means,
Conventionally, bandpass filters using analog circuits have been mainly used.However, if a plurality of frequency components are to be simultaneously extracted, the same number of analog circuits as the number of types of frequency components to be extracted are required. Increasing the scale and complicating the circuit adjustment process are inevitable. Therefore, in this embodiment, a knock signal is AD-converted, and a frequency component is extracted from the AD conversion result by a digital filter. Since the frequency shown in FIG. 6 varies depending on the type of engine, the shape of the combustion chamber, the bore diameter and the like, the adjustment can be simplified by configuring the frequency extracting means 17 with a digital filter as in this embodiment.

As the digital filter, there are a non-regression type filter shown in FIG. 2 and a regression type filter shown in FIG. 3. In this embodiment, either digital filter may be used. Furthermore, as shown in FIG. 4, the digital filter may be a filter that calculates the difference between the output of the all-band pass filter and the output of the band elimination filter for each characteristic frequency. In FIG. 4, the delay circuit functions as an all-pass filter. Then, filters are prepared by the number of frequency components to be extracted, and necessary types of frequency components are simultaneously extracted.

The sampling interval τ _s at which the AD conversion is performed is the reciprocal of a frequency that is at least twice the maximum frequency to be extracted according to the sampling theorem, and in the example of FIG. .1 kHz,
The interval τ _s may satisfy the following equation.

[0021]

(Equation 1)

The coefficients ai and bi of the non-regression type filter shown in FIG. 2 and the regression type filter shown in FIG. 3 are prepared in advance in accordance with the digital filter design procedure.

On the other hand, a fast Fourier transform is known as a frequency analyzing means. In this case, a number of 2 n samples is required. For this reason, the section to be subjected to frequency analysis, that is, the sampling section is a double or １ ／ discrete section and is a discontinuous section.

In order to solve this problem, there is a method in which each AD sample value is multiplied by a Fourier coefficient, regardless of the number of 2n samplings. One embodiment is shown in FIG.
That is, the frequency analysis means based on the Fourier series shown in FIG. 7 sets the frequency to be extracted to f, sets the Fourier coefficient ci as follows, and thereby changes the coefficient every time the number of times of sampling i changes to change the frequency component. Extract it.

[0025]

(Equation 2)

At this time, as is evident from FIG. 7, in this embodiment, the correction means 40 is provided, and the combustion in the cylinder according to the operating state of the engine, for example, the fuel injection width _TP and the rotational speed N is performed. Estimate the temperature and calculate the frequency f
Is changed so that an appropriate frequency component can always be extracted.

By the way, according to the present invention, the non-regression type digital filter of FIG. 2 and the regression type digital filter of FIG.
Alternatively, each of the frequency components S _{1 to} S ₅ by either the filter of FIG. 4 or the frequency analysis means based on the Fourier series of FIG.
In the extraction processing of, at least three types of frequency components are further extracted, each centering on each characteristic frequency, and the frequency at which the amplitude level has the maximum value is determined as the corresponding characteristic frequency component. The main feature is that the center frequency of the filter is moved based on the relationship between this and the remaining at least two kinds of frequency components so that the filter is always tuned to each characteristic frequency. , Below,
This will be described with reference to FIG.

In FIG. 1, except for the knock sensor and the sampling process, processes 111 to 114 correspond to the non-regression type digital filter of FIG. 2, the regression type digital filter of FIG. 3, the filter of FIG. 4, or the Fourier series of FIG. Corresponding to each frequency component extraction process in the frequency analysis means.

First, as shown in FIG. 1, three types of frequency analysis processing 111, 112, and 113 are performed.
, S ₁₁ = f ₁ −Δf, S ₁₂ = f ₁ , S ₁₃ = f for each feature frequency, as shown in the spectrum diagram at the lower left of the figure.
Three frequency components of ₁ + Δf are extracted. Here, f ₁ is the center frequency adjusted to the characteristic frequency f ₀ , and the value of Δf may be set to about several hundred Hz, depending on the value of the center frequency f ₁ .

[0030] Then, Now, the frequency components S ₁₃ is, if indicated the maximum level among these three frequency components, since at this time indicates that the shifting to the higher by at least Δf is characteristic frequency , Center frequency f ₁
Is increased by Δf. Since Conversely, the frequency components S ₁₁ is, if indicated the maximum level among these three frequency components, in this case indicates that is shifted to lower by at least Δf characteristic frequency, this time around to lower the frequency f ₁ only Δf.

[0031] Thus, when the amplitude level of the frequency components S ₁₂ was maximum, towards the amplitude level greater in both sides of the frequency components S ₁₁ and the frequency components S _13, the center frequency f
₁ is shifted by Δf × x (x = 0 to 1). Accordingly, the above processing is as shown in processing 114 in FIG. 1. At this time, as a specific method for shifting the frequency, in the above-described digital filter, each coefficient ai, bi, ci is set. It only needs to be changed and can be realized very easily.

When the frequency components S _{1 to} S ₅ have been extracted in this manner, the processing by the knock determination means shown in FIG. 8 is performed. First, in processes 80 and 81, the frequency components S _{1 to} S ₅ of the knock vibration are the average values S ₁ to S ₅ of the frequency components S _{1 to} S ₅ when it is determined that there is no knock so far.
S ₅ (indicated by superscript bar in the following equation and FIG. 8)
, And their ratios or differences (zero if negative) are determined, and their values are set to SN ₁ to SN ₅ .

Next, at step 82, these values SN ₁
ＳＮSN ₅ , the sum of them, or the sum of the top m when they are arranged in descending order of value, are taken as the knock index IK
Output as Then, the knock index IK is calculated as processing 8
When the knock index IK signal corresponding to the upper y% of the cumulative frequency distribution of the knock index IK signal shown in the lower right of FIG. It is determined to have occurred. Where
The cumulative frequency distribution at the lower right of the graph of> shows that the level of the large index is 0%, and all the index levels are added to the low index level.

Next, if it is determined that no knocking, in the process 84 is performed to update the average value S ₁ ~ S ₅ frequency components S ₁ to S ₅ in the following manner.

[0035]

(Equation 3)

Here, K is a coefficient for obtaining a moving average, and a numerical value between K = 4 and 64 is used. Next, a frequency component extraction timing process will be described. First, FIG. 9 shows a configuration necessary for this processing. The knock signal from the knock sensor 15, the reference signal from the reference angle sensor 11, and the position signal from the position angle sensor 12 are input, and the position Count the rise of the signal,
Counter 10 cleared at rising of reference signal
0, a compare register 101 for comparing the count value of the counter 100, an interrupt controller 102 for requesting an interrupt to the CPU 103 when the count value of the counter 100 matches the count value of the compare register 101, and an AD conversion for AD converting the knock signal. And a vessel 104.

Next, the processing operation will be described with reference to the flowchart of FIG. 10 and the timing chart of FIG. 11. First, as shown in FIG. 10A, the CPU 103 is interrupted at the rise of the reference signal Ref. AD
The angle θ _{K at} which the conversion operation by the converter 104 should be started is set in the conveyor register 101.

The counter 100 has a reference signal Ref.
Is cleared at the rising edge of, and the counting of the position signal Pos starts from zero. And the count value of the counter 100 is θ _K
If the values match, the process by the coincidence interrupt shown in FIG. 10B is started, in which the AD conversion by the AD converter 104 is started, and at the same time, the AD conversion interrupt is permitted. Then, the end angle θ _N of the AD conversion is set in the compare register 101.

Further, each time the A / D conversion is completed, an A / D conversion interrupt request is issued to the CPU, and the processing of FIG. 10C is started, and the product-sum operation of the digital filter is performed within the A / D conversion interrupt. This product-sum operation is performed by converting the AD conversion value ADi and the coefficient a
This is a multiplication with i, bi, ci, and the like and an operation for obtaining the sum thereof, which is performed until the crank angle reaches θ _N.

When the count value of the counter 100 counting the position signal Pos coincides with θ _N , a coincidence interrupt request is issued again, in which the AD conversion interrupt is stopped, and the product sum up to that time is stopped. The calculated value is set to each of the frequency components S _{1 to} S ₅ .

Then, as shown in FIG. 8, the ratio or difference SN ₁ -S between these frequency components S ₁ -S ₅ and the average value.
Seeking N _5, it is to calculate the knock index IK from these values.

After that, according to the processing 83 of FIG. 8, when the knock index IK is larger than the level SL in the sensory test, it is determined that there is knock, and when it is smaller, it is determined that there is no knock. As shown in FIG. 11, the signal is output until the next rising of the reference signal Ref.

Therefore, the engine control unit 1 inputs the presence / absence of the knock at the rising of the next reference signal Ref. Perform the control.

As described above, according to this embodiment, occurrence of knocking can always be reliably detected, so that accurate knocking avoidance control can be performed, and engine performance can be sufficiently improved.

Further, according to this embodiment, since the information contained in the output of the vibration sensor can be effectively used, control can be performed so that the engine output and the fuel efficiency are optimized.

[0046]

According to the present invention, a plurality of frequency components can be analyzed simultaneously according to the vibration mode in the cylinder, and the analysis result becomes the signal intensity included in the analysis section, so that hardware is required. Since it can be simplified and the degree of freedom of frequency selection increases, there is an effect that even if the specifications of the target engine are changed, it can be handled.

Further, according to the present invention, the analysis frequency is automatically fine-tuned according to the vibration mode in the cylinder, and automatically tracks the change in the analysis frequency. Nevertheless, knock can always be reliably detected, and highly accurate knock control can be obtained.

Furthermore, the sampling range is not limited to the number of 2 n samples called the fast Fourier transform, and the sampling range can be freely selected. Therefore, the frequency analysis section can be freely set according to the engine speed, and the knocking detection can be performed. Accuracy can be sufficiently improved.

[Brief description of the drawings]

FIG. 1 is a block diagram showing frequency analysis means in an embodiment of a knocking detection device according to the present invention.

FIG. 2 shows a first example of a filter means in one embodiment of the present invention.
It is a block diagram which shows the example of.

FIG. 3 shows a second example of the filter means in one embodiment of the present invention.
It is a block diagram which shows the example of.

FIG. 4 shows a third example of the filter means in one embodiment of the present invention.
It is a block diagram which shows the example of.

FIG. 5 is a configuration diagram of an engine control device to which an embodiment of a knocking detection device according to the present invention is applied.

FIG. 6 is an explanatory diagram of a knocking frequency mode.

FIG. 7 shows a fourth embodiment of the filter means in one embodiment of the present invention.
It is a block diagram which shows the example of.

FIG. 8 is a block diagram of a knocking determination unit according to one embodiment of the present invention.

FIG. 9 is a block diagram of a timing control unit in one embodiment of the present invention.

FIG. 10 is a flowchart illustrating a timing control operation in one embodiment of the present invention.

FIG. 11 is a timing chart illustrating a timing control operation in one embodiment of the present invention.

[Explanation of symbols]

 Reference Signs List 1 engine control unit 2 engine cylinder 3 intake pipe 4 exhaust pipe 5 throttle valve 6 fuel injection valve 7 spark plug 8 ignition coil 9 distributor 10 intake air flow meter 11 reference angle sensor 12 position angle sensor 13 throttle valve opening sensor 14 empty Fuel ratio sensor 15A Knock vibration sensor 15B Seat pressure sensor 15C In-cylinder pressure sensor

──────────────────────────────────────────────────の Continuation of the front page (51) Int.Cl. ⁷ identification code FI G01M 15/00 G01M 15/00 A (58) Field surveyed (Int.Cl. ⁷ , DB name) G01H 17/00 F02B 77 / 08 F02D 45/00 358 F02D 45/00 368 G01M 15/00

Claims (2)

(57) [Claims] 1. Frequency component extracting means for independently extracting at least two predetermined frequency components included in an output signal of a knock sensor for detecting at least one of engine vibration and cylinder pressure vibration, and frequency component extracting means. And a knock index calculating means for calculating a knock index from a calculation result of at least two outputs given from the above, and making the at least two frequency components coincide with resonance frequency components determined by respective resonance modes appearing in the cylinder of the engine. A knocking detection device that determines knocking occurrence from the knocking index, wherein at least 3 for each of the at least two frequency components.
Frequency analysis means including a digital filter for extracting signals having frequencies close to each other, and changing a center frequency of the frequency analysis means according to a frequency of a signal having a maximum amplitude level among the at least three kinds of signals. Processing means, and correction means for changing the frequency of the digital filter according to the operating state of the engine, wherein the center frequency of the frequency analysis means is automatically tracked by the change in the frequency of the resonance frequency component, and the frequency by the digital filter is A knocking detection device, wherein an analysis section is configured to be changed according to an operation state of an engine. 2. The invention according to claim 1 , wherein said correction means includes means for calculating a sound speed in an atmosphere in a cylinder from an operation state of an engine, and said digital filter is operated in accordance with a result of calculation of the sound speed by said means. A knocking detection device characterized in that the center frequency is changed.