JP2002047994A - Knock control system of internal combustion engine - Google Patents

Abstract

PROBLEM TO BE SOLVED: To accurately judge whether a knock occurs or not by properly setting a knock judgement level for a frequency band for detecting a knock and to control a knock. SOLUTION: A signal of a frequency band peculiar to a knock is filtered and a peak value effective for detecting a knock is detected from a vibration waveform signal detected by a knock sensor 5 and generated in an internal combustion engine 1 in a representative value detection circuit 9. Here, an upper limit value and a lower limit value are also changed so that standard deviation as a parameter obtained from the peak value for setting a knock judgement level in accordance with switching of filtering frequency bands in the representative value detection circuit 9 based on an operation condition of the internal combustion engine 1 is within the optimum scope. The knock judgement level set in this way is compared with the peak value to judge whether a knock occurs in the internal combustion engine 1 or not accurately. An operation condition of the internal combustion engine 1 can be properly controlled according to the thus obtained results of judgement.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a knock control device for an internal combustion engine that detects the occurrence of knock in the internal combustion engine and controls the operating state of the internal combustion engine.

[0002]

2. Description of the Related Art Conventionally, as prior art documents relating to a knock control device for an internal combustion engine, those disclosed in Japanese Patent No. 2730215 and Japanese Patent Application Laid-Open No. 5-306645 are known. Among them, Japanese Patent No. 2730215 discloses a technique for creating a knock determination level based on a substantial standard deviation or the like in a frequency distribution of knock intensity values in order to realize optimal knock detection. In addition, there is disclosed a technique in which an upper limit value and a lower limit value are provided for the standard deviation to prevent a decrease in the standard deviation when the knock intensity value is minimum and a divergence of the standard deviation when the knock intensity value is maximum. Japanese Patent Application Laid-Open No. Hei 5-306645 discloses a technique for variably selecting a center frequency of a knock detection filter to improve knock detection.

[0003]

By the way, in the above-mentioned system, when the knock frequency in the internal combustion engine changes, the vibration characteristic of the knock sensor also changes greatly, so that the standard deviation of the knock intensity value also changes greatly. For this reason,
Since the knock determination level changes, the presence or absence of knock cannot be accurately determined.

Accordingly, the present invention has been made to solve such a problem, and an internal combustion engine capable of accurately judging the presence / absence of knocking and performing knock control by appropriately setting a knock judging level for a frequency band of knock detection. It is an object to provide an engine knock control device.

[0005]

According to the knock control apparatus for an internal combustion engine of the present invention, a representative value detecting means detects a frequency band specific to knock from a vibration waveform signal generated in the internal combustion engine detected by the signal detecting means. The signal is filtered to find a representative value useful for knock detection. Here, based on the operating state of the internal combustion engine detected by the operating state detecting means, a parameter for setting the knock determination level in accordance with the switching of the filtering frequency band in the representative value detecting means by the frequency switching means is a limit value changing means. The upper and lower limits are also changed so as to be within the optimum range. By comparing the knock determination level thus set and the representative value with the knock determination means, it is possible to accurately determine whether knock has occurred in the internal combustion engine. The knock control means appropriately controls the operating state of the internal combustion engine based on the determination result obtained in this manner.

In the knock control apparatus for an internal combustion engine according to the present invention, the knock determination level is set using a standard deviation value as a parameter obtained from a representative value in a knock-specific frequency band, so that knock generation in the internal combustion engine is performed. Is accurately determined.

In the knock control device for an internal combustion engine according to the third aspect, the upper limit value and the lower limit value of the standard deviation value are set so as to be within the optimum range for each center frequency of the knock-specific frequency band. According to the knock determination level set using such a standard deviation value, the reliability is high and the presence or absence of knock in the internal combustion engine is accurately determined.

[0008]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of the present invention will be described below based on examples.

FIG. 1 is a schematic diagram showing a configuration of an internal combustion engine to which a knock control device for an internal combustion engine according to one embodiment of the present invention is applied.

In FIG. 1, reference numeral 1 denotes a four-cylinder four-cycle internal combustion engine, and intake air is supplied to the internal combustion engine 1 from an intake passage 2 through an air cleaner and an air flow meter (not shown). The amount of intake air supplied to the internal combustion engine 1 is adjusted by the throttle valve 3. Reference numeral 4 denotes a reference angle sensor 4a for detecting a reference crank angle of the internal combustion engine 1 (for example, a compression top dead center of each cylinder) and a crank angle sensor 4b for generating an output signal at every constant crank angle of the internal combustion engine 1. It is a built-in distributor. And EC which will be described later
U (Electronic Control Unit) 10
The engine speed NE is calculated based on the signal from the crank angle sensor 4b.

Reference numeral 5 denotes a knock sensor for detecting vibration of a cylinder block corresponding to a knocking phenomenon of the internal combustion engine 1 by a piezoelectric element type (piezo element type), an electromagnetic type (magnet, coil) or the like, and 6 denotes a cooling of the internal combustion engine 1. This is a water temperature sensor that generates a signal according to the water temperature. 7 is the ECU 1
An injector (fuel injection valve) that supplies fuel to the intake manifold based on a control signal from 0. Reference numeral 8 denotes a spark plug that ignites and ignites a mixture in a cylinder of the internal combustion engine 1 with a high voltage distributed from the distributor 4.

Reference numeral 9 denotes a representative value detection circuit for detecting a representative value from an output signal (knock sensor signal) from the knock sensor 5, and as shown in FIG. 2, selects only a knock frequency component from the knock sensor signal. A filter circuit 9a for extracting the signal, an amplifier circuit 9b for amplifying a signal from the filter circuit 9a, and a peak hold circuit 9 for detecting a peak value VP of a knock sensor signal within a predetermined crank angle as a representative value.
c. Here, the predetermined crank angle is, for example, a fixed period (knock determination period) after the ignition output in order to prevent erroneous detection of knock due to noise because knock occurs only in the combustion pressure in the present embodiment. .

The ECU 10 includes a CPU 11 serving as a central processing unit for executing various known arithmetic processing, a ROM 12 storing a control program, and a RAM 1 storing various data.
3. A logical operation circuit mainly composed of a B / U (backup) RAM 14, an A / D converter 15 for analog-digital conversion, etc., and an input port 16 for inputting detection signals from various sensors and control signals for various actuators. Are connected to each other via a bus 18 to an output port 17 or the like for outputting Then, the ECU 10 controls the ignition timing of the ignition plug 8, the fuel injection amount from the injector 7, and the like according to various sensor signals and the knock occurrence state.

Next, an ECU used in a knock control device for an internal combustion engine according to an embodiment of the present invention.
The processing procedure of knock detection / judgment control by the CPU 11 in the CPU 10 will be described with reference to flowcharts of FIGS. 3, 6, 8, 10, and 12. This knock detection / determination control routine is repeatedly executed for each combustion cycle of the internal combustion engine 1, and includes a fall of a gate signal output from the ECU 10 to the peak hold circuit 9c.
That is, the CPU 11 starts the execution immediately after reading the peak value VP.

In FIG. 3, in step S101, a peak value VP detected by the peak hold circuit 8c is read as a representative value. Next, in step S102, the peak value VP read in step S101 is logarithmically converted by the following equation (1) as logarithmic conversion processing, and the frequency distribution of the peak value VP is corrected so as to be a normal distribution. . Here, A and a are constants (details will be described later), and A = 64 / lo
g4, a = 4. Further, a value obtained by logarithmically converting the peak value VP is expressed as lVP.

[0016]

## EQU1 ## lVP = A.times.log (VP / a) (1)

Next, the process proceeds to step S103, in which a knock determination level VJL is set as a knock determination level setting process (details will be described later). Next, the process proceeds to step S104, and it is determined that knock is present when lVP ≧ VJL as a knock determination process (details will be described later). Based on this determination result, the ignition timing and the like are controlled in a knock control region (for example, a high load region). Next, the process proceeds to step S105, and it is determined whether the operating condition of the internal combustion engine satisfies a specific condition. The specific conditions are, for example, that the load of the internal combustion engine is equal to or higher than a predetermined value, the engine speed NE is within a predetermined range, the rate of change of the load of the internal combustion engine is equal to or lower than a predetermined value, and the median value VME
D is defined by one or more of the conditions within a predetermined range.

If the determination condition of step S105 is satisfied, that is, if the operating condition of the internal combustion engine satisfies the specific condition, the process proceeds to step S106, and the standard deviation is updated to a value substantially equal to the standard deviation of the lVP distribution (normal distribution). The SGM is updated (that is, the actual standard deviation value, hereinafter simply referred to as “standard deviation”) (details will be described later). On the other hand, when the determination condition of step S105 is not satisfied, that is, when the operating condition of the internal combustion engine does not satisfy the specific condition, step S1
06 is skipped. Next, the processing shifts to step S107, in which the median value update processing is performed with the 50% point in the cumulative% point,
That is, the median value VMED of the lVP distribution is updated (details will be described later), and this routine ends.

Next, the logarithmic conversion processing in step S102 will be described in detail.

In the above equation (1), A = 64 / log
4, a = 4, but it is not always necessary to set this value. However, considering the current state of the art, the constants A, a
Is optimally set to about the above value. The meaning of these constants A and a divides the entire lognormal distribution (see FIG. 4) of the peak value VP when no knock occurs at all (the ratio of a large output to a small output is about 4) by 64. Also, 4
A 10-bit A / D converted peak value VP is compressed by the A / D converter 15 into 8-bit data so that it can be easily handled by a 8-bit or 8-bit CPU.

Next, an expression that satisfies the above two conditions is derived.

First, to convert the lognormal distribution as shown in FIG. 4 to a normal distribution as shown in FIG.
(3) is required.

[0023]

## EQU2 ## 1VL = A.times.log (VL / a) (2)

[0024]

## EQU3 ## lVH = A.times.log (VH / a) (3)

Here, in order to divide the entire distribution into 64, the conditions of the following equations (4) and (5) must be satisfied.

[0026]

## EQU4 ## lVH-lVL = 64 (4)

[0027]

VH = 4 × VL (5)

Therefore, the following equation (6) is derived from the above equations (2) to (5).

[0029]

A × log4 = 64 ∴A = 64 / log4 (6)

The maximum value 2 ¹⁰ of the 10-bit A / D conversion value of VH is converted to 1VH, and the maximum value 2 ^{8 of the 8-} bit data is converted.
The following equation (7) is derived.

[0031]

2 ⁸ = A × log 2 ¹⁰ / a ∴a = 1024/1 × p (256 / A) (7)

Therefore, the logarithmic conversion equation that satisfies the above condition is given by the following equation (8).

[0033]

[Expression 8] 1V = (64 / log4) × (logV / 4) (8)

In the current state of the art, the CPU 11 directly
Since logarithmic conversion takes time and cannot be applied in actual control, the peak value V
The logarithmic conversion value lVP corresponding to P may be calculated, stored in the ROM 12, and sequentially read. Further, if the peak value VP is more than 2 ^8, a quarter of the peak value VP
Can be obtained by reading the logarithmically converted value lVP corresponding to the value from the ROM 12 and adding 64 to the value, so that the data stored in the ROM 12 is 2 ⁸ (= 256) bytes Only needs to be done.

The knock determination level setting process in step S103 in FIG. 3 will be described with reference to the flowchart shown in FIG.

In FIG. 6, in step S201, the engine speed NE of the internal combustion engine is logarithmically converted. The method of logarithmic conversion is the same as step S102 in FIG. Next, the process proceeds to step S202, where the logarithmic conversion value 1NE of the engine speed NE at the current control timing is set.
Is obtained from the logarithmically converted value 1NEo of the engine speed NE at the previous control timing. Next, the routine proceeds to step S203, where the logarithmic conversion value lNE of the engine speed NE at the current control timing is stored as the logarithm conversion value lNEo of the previous engine rotation speed NE at the next control timing. Next, step S20
Then, the average value VMALL of the logarithmic conversion value lVP of the peak value VP in all the cylinders is corrected by the following equation (9).
Here, β is a constant of about 2, and the average value VMALL is shown in FIG.
Is updated at each control timing in step S107 (details will be described later).

[0037]

[Equation 9] VMALL ← VMALL + β × dlNE (9)

Next, the flow shifts to step S205, where the median value VMED is corrected by the following equation (10). Where dVME
D is the deviation between the median value VMED and the average value VMALL for each cylinder. This deviation dVMED is updated at each control timing in step S107 in FIG.

[0039]

[Equation 10] VMED ← VMALL + dVMED (10)

Then, the flow shifts to step S206, where the knock determination level VJL is set by the following equation (11), and this routine ends. Here, u is a value set according to the engine speed NE (details will be described later).

[0041]

[Expression 11] VJL ← u × SGM + VMED (11)

Next, the correction of the average value VMALL in step S204 of FIG. 6 will be described. When the inventors conducted experiments on various internal combustion engines, it was found that the median value VMED of the peak value VP was given by the following equation (12). Here, V0 and β are constants determined by the internal combustion engine, the cylinder, the knock sensor, and the like, and β is a value of about 1.5 to 2.5.

[0043]

(Equation 12) ... (12)

Now, when the engine speed NE increases at a rate α [rpm
/ Sec], the rate of increase dV / dn of the median value VMED per combustion cycle is given by the following equation (1).
It is represented by 3).

[0045]

(Equation 13) ... (13)

Here, when approximation by β = 2, dV / dn
= 240αV0. That is, by adding 240αV0 to the average value VMALL for each combustion cycle, it is possible to prevent a delay in following the average value VMALL during transition.
However, since V0 varies greatly depending on the internal combustion engine, the cylinder, the knock sensor, and the like, there is no added value that satisfies all of these combinations. Therefore, the average value V
It is difficult to properly correct MALL. The influence of V0 can be eliminated by logarithmic conversion of the peak value VP and the engine speed NE. More specifically, first, both sides of the above equation (12) are logarithmically converted to obtain the following equation (14).

[0047]

[Equation 14] ... (14)

Here, if a value obtained by logarithmically converting x is expressed as lx, it is expressed by the following equation (15).

[0049]

[Expression 15] lVMED = A × logV0 + β × 1NE (15)

In the above equation (15), the first term on the right-hand side is a constant term. Therefore, differentiating both sides gives the following equation (16).

[0051]

DlVMED / dn = β × dlNE / dn (16)

Therefore, dlNE / dn is detected for each combustion cycle or for each ignition timing, and β × dlN
By adding E / dn, the average value VMALL can be corrected ignoring V0.

Next, the knock determination level setting processing in step S206 in FIG. 6 will be described.

Now, the specification of the amount of advance and retard is set to 1 for one knock.
[° CA (Crank Angle)] It is assumed that the retard angle is advanced by 1 [° CA] every second. In this specification of the amount of advance and retard, when no knock occurs, the knock determination frequency at which the advance and retard are stabilized is 1 [times / sec]. When the knock determination level VJL is set so that the knock determination frequency is 1 [times / sec], the retard amount R converges to a constant value, so that the ignition timing is controlled to be substantially constant.

However, the maximum torque of the internal combustion engine is obtained in a region where slight knock occurs. Therefore, it is necessary to control the ignition timing to always advance in a state where knock does not occur. Therefore, the knock determination frequency may be set to be smaller than the knock determination frequency at which the advance / retard angle is stable as described above.

Now, in a four-cylinder internal combustion engine, if the knock determination frequency is 1/4 [times / sec], the knock determination probability P
K is expressed by the following equation (17).

[0057]

## EQU17 ## PK = (120 / NE) .times. (1/4) .times. (1/4) (17)

When the engine speed NE is 2000 [rpm]
m], the knock determination probability PK = 0.0375,
The u value for this knock determination probability PK is calculated from the normal distribution by u
= 2.67. Also, the engine speed N
If E is 4000 [rpm], PK = 0.00187
5, which can lead to u = 2.90. FIG. 7 shows a characteristic diagram of the engine speed NE and the u value. Here, T is a knock detection cycle. As is clear from the characteristic diagram of FIG. 7, when the knock detection frequency is constant (T = constant), the u value increases as the engine rotational speed NE increases.

Normally, the specifications of the amount of advance and retard are constant.
Knock detection frequency can be uniquely set. Therefore, the u value corresponding to the engine speed NE may be obtained in advance from the normal distribution table and stored in the ROM 12, and the u value corresponding to the engine speed NE may be sequentially read.

Further, since knocking is unlikely to occur at a light load, the load on the internal combustion engine 1 may be detected, and the u value may be increased at a light load. In the present embodiment, the u value is set to a predetermined knock detection frequency (for example, T = 4.0).
[Sec]), but in order to control the ignition timing to always advance when knock does not occur, for example, in the specification of the advance / retard amount described above, The u value may be set to a value larger than the characteristic of T = 1.0 [sec] shown in FIG.

Next, the knock determination processing in step S104 of FIG. 3 will be described in detail with reference to the flowchart shown in FIG.

In FIG. 8, first, at step S301
It is determined whether the value lVP obtained by logarithmically converting the peak value VP is equal to or higher than the knock determination level VJL. Step S30
When the inequality of 1VP ≧ VJL is satisfied in 1 and it is determined that knock has occurred, the process proceeds to step S302,
It is determined whether the logarithm conversion value lVP satisfies the following equation (18).

[0063]

[Expression 18] lVP ≧ VJL + SGM (18)

If the condition of step S302 is not satisfied,
That is, when the inequality expression (18) is not satisfied, the process proceeds to step S303, where a predetermined value ΔR is added to the retard amount R, and the retard amount R is set to increase by the predetermined value ΔR. Here, ΔR is the specification of the amount of advance and retard, and ΔR
= 1 [° CA]. On the other hand, when the determination condition is satisfied in step S302, that is, when the inequality expression (18) is satisfied, the process proceeds to step S304, and it is determined whether the logarithmically converted value lVP satisfies the following expression (19).

[0065]

[Expression 19] lVP ≧ VJL + 2 × SGM (19)

If the condition of step S304 is not satisfied,
That is, when the inequality expression (19) is not satisfied, the process proceeds to step S305, and the retard amount R is set to a predetermined value (2 × ΔR).
Is added so that the retard amount R is increased by a predetermined value (2 × ΔR). On the other hand, when the determination condition of step S304 is satisfied, that is, when the inequality expression (19) is satisfied, the process proceeds to step S306, and the retard amount R is set to a predetermined value (3 ×
ΔR) is added and the retard amount R is set to increase by a predetermined value (3 × ΔR). After the processing of step S303, step S305 or step S306, the process proceeds to step S307, and it is determined whether or not the retard amount R exceeds the upper limit value Rmax. When the determination condition of step S307 is satisfied, that is, when the retard amount R is larger than the upper limit value Rmax, the process proceeds to step S308, and after the retard amount R is set to the upper limit value Rmax as a guard process, the present routine ends. .

On the other hand, when it is determined in step S301 that the inequality of lVP ≧ VJL does not hold and no knock has occurred, or in step S307, the retard amount R is set to the upper limit value R.
If it is smaller than max, the process moves to step S309. In step S309, it is determined whether a predetermined time Tad has elapsed. Here, the predetermined time Tad is a time required for the advance angle, and is, for example, 1 [sec]. Step S3
When the determination condition of 09 is satisfied, that is, when the predetermined time Tad has elapsed, the process proceeds to step S310, and the predetermined value ΔR is subtracted from the retard amount R, and the retard amount R is set to be smaller by the predetermined value ΔR. . Next, the process proceeds to step S311, and it is determined whether or not the retard amount R is less than the lower limit value Rmin. When the determination condition of step S311 is satisfied, that is, when the retard amount R is smaller than the lower limit value Rmin, step S31 is performed.
2 and the retard amount R is set to the lower limit value Rmin as a guard process.
After this is set, this routine ends. On the other hand, the determination condition of step S309 is not satisfied, that is, the predetermined time T
When ad has not passed yet, or in step S311
Is not satisfied, that is, the retard amount R is equal to the lower limit value Rmin.
When it is larger than the above, this routine is terminated without doing anything.

Here, considering the controllability, the knock is 1-2.
It is preferable to set the retardation amount R to be larger when it occurs in one cylinder in a concentrated manner than when it occurs in a plurality of cylinders. For example, the cylinder in which the knock has occurred is stored, and if the knock occurs continuously in the same cylinder, the retard amount R is further reduced by a predetermined value (2
× ΔR).

Next, in step S302 in FIG.
How the retard amount R is changed depending on whether or not the logarithmic conversion value lVP is larger than the value obtained by adding the standard deviation SGM to the knock determination level VJL will be described with reference to FIG. FIG. 9 (a)
9B, the distribution of the logarithmically converted value lVP when no knock occurs is a normal distribution as shown by a solid line without knock, and when the knock occurs (no knock occurs). When the combustion cycle and the combustion cycle in which knock occurs are mixed, the distribution is distorted on the high output side as indicated by the broken line because knock is present.

Also, as shown in FIG. 9 (a), even when the knocking occurs in the same knocking state, as shown in FIG. 9 (a), compared with the case where the standard deviation SGM of the logarithmically converted value lVP is small, no knocking occurs. As shown in (b), the standard deviation SGM
Is larger when is larger. However, the probability of the broken portion is the same. Therefore, assuming that the determination level for determining the knock intensity is VJL'1 = VJL + k (k is a constant) with respect to the basic knock determination level VJL, the logarithmic conversion value lVP exceeds the knock determination level VJL. There is a problem that the probabilities are different in spite of the same knock occurrence state.

Here, in FIGS. 9 (a) and 9 (b), since we are talking on the logarithmic axis, the constant k is added, but on the real axis, the knock determination level VJL Is multiplied by a constant k. That is, since the knock strength determination level VJL'1 is also a constant multiple of the basic determination level, accurate knock strength determination cannot be performed even if the basic determination level is an optimum value. On the other hand, as shown in FIGS. 9C and 9D, if the knock intensity determination level VJL'2 = VJL + SGM is set based on the standard deviation SGM, the knock intensity determination level according to the distribution shape is obtained. VJL'2 can be set. That is, the probability of exceeding the knock intensity determination level VJL'2 can be made the same for all distribution shapes.

Next, the standard deviation updating process in step S106 in FIG. 3 will be described with reference to the flowchart shown in FIG.

In FIG. 10, first, at step S401
Then, the center frequency f0 of the knock detection frequency band is set. The center frequency f0 is appropriately switched and set using a well-known SCF (switched capacitor filter), a plurality of BPFs (bandpass filters), etc., which constitute the filter circuit 9a in the representative value detection circuit 9. For example, when the engine speed NE of the internal combustion engine 1 is 2000 [rpm
m], 2000 to 4000 [rpm] or 400
0 [rpm] depending on whether the center frequency f0
Are set to X, Y, and Z, respectively. Next, step S4
02, the upper limit SGMmax and the lower limit SGMmin of the standard deviation SGM with respect to each center frequency f0 set in step S401 are set as guard processing.

Then, the flow shifts to step S403, where it is determined whether or not the logarithm conversion value lVP is within the range between (VMED-SGM) and VMED. The determination condition of step S403 is satisfied, that is, the logarithm conversion value lVP is (VMED-SGM) and VME
If it is within the range of D, the process proceeds to step S404, and the standard deviation SGM is updated by the following equation (20). Here, dsgm is an update amount of the standard deviation SGM, and is set according to a change in the operating state of the internal combustion engine. For example, a deviation dlN of a logarithmic conversion value 1NE of the engine speed NE.
When E is large, the update amount dsgm is also set large. First, even when the change in the load of the internal combustion engine is large, the update amount dsg
m is set large.

[0075]

SGM ← SGM−2 × dsgm (20)

On the other hand, the judgment condition of step S403 is not satisfied, that is, the logarithm conversion value lVP is (VMED-SGM) and V
If it is not within the range of MED, the flow shifts to step S405, and the standard deviation SGM is updated by the following equation (21).

[0077]

SGM ← SGM + dsgm (21)

Step S404 or step S405
After the processing in step S406, the flow shifts to step S406, and it is determined whether the standard deviation SGM exceeds the upper limit SGMmax.
The determination condition of step S406 is satisfied, that is, the standard deviation S
If the GM is larger than the upper limit SGMmax and is larger, step S4
07, the standard deviation SGM is set to the upper limit SGMmax as a guard process. On the other hand, the determination condition of step S406 is not satisfied, that is, the standard deviation SGM is equal to the upper limit value SGMmax.
If smaller, step S407 is skipped.

Next, the flow shifts to step S408, where it is determined whether or not the standard deviation SGM is smaller than the lower limit SGMmin. The determination condition of step S408 is satisfied, that is, the standard deviation SGM
Is smaller than the lower limit value SGMmin, step S40
9 and the standard deviation SGM is set to the lower limit S as guard processing.
Set to GMmin and end this routine. On the other hand, the determination condition of step S408 is not satisfied, that is, the standard deviation SGM
Is greater than or equal to the lower limit value SGMmin, step S40
9 is skipped, and this routine ends.

Here, the fact that the standard deviation SGM is obtained by the above-described processing will be described.

In FIG. 11A, the logarithmic conversion value lVP
Let Ps be the probability that the standard deviation SGM is within the range (hatched portion) from (VMED-SGM) to VMED, at the point where the standard deviation SGM is stabilized by the processing of the flowchart shown in FIG. Become.

[0082]

-Ps × 2 × dsgm + (1−Ps) × dsgm = 0∴Ps = 1/3 (22)

When the u value corresponding to the standard deviation SGM is obtained from the normal distribution table, u ≒ 0.97, and the standard deviation (u
= 1.0). Similarly, FIG.
In (b), the logarithm conversion value lVP is (VMED-SGM)
Assuming that the probability of being within the range of (VMED + SGM) (shaded area) is Ps, the point where the standard deviation SGM is stabilized by the processing of the above-described flowchart shown in FIG. 10 is represented by the following equation (23).

[0084]

-Ps × dsgm + (1-Ps) × 2 × dsgm = 0∴Ps = 2/3 (23)

When the u value corresponding to the standard deviation SGM is obtained from the normal distribution table, the value becomes substantially equal to the standard deviation. Therefore, the standard deviation SGM can be easily obtained by the processing of the flowchart shown in FIG. Also,
The method is not limited to the method described above, and for example, step S404 in FIG. 10 may be changed to the following equation (24).

[0086]

SGM ← SGM−3 × dsgm (24)

However, in this case, Ps = １ ／, and
u = 0.67, and the standard deviation SGM obtained by the above equation (24) is 0.67 times the actual standard deviation.
Therefore, in consideration of the balance between the increase and decrease of the standard deviation SGM or the ease of post-processing, the processing shown in FIG. 10 is more preferable.

Next, the median value updating process in step S107 in FIG. 3 will be described with reference to the flowchart shown in FIG.

In FIG. 12, first, at step S501
, It is determined whether the logarithmically converted value lVP exceeds the median value VMED. When the determination condition of step S501 is satisfied, that is, when the logarithmic conversion value lVP exceeds the median value VMED and is large, the process proceeds to step S502, and the median value deviation dVMED.
Is updated by the following equation (25). Here, dVm is the update amount of the deviation dVMED.

[0090]

[Expression 25] dVMED ← dVMED + dVm (25)

On the other hand, if the determination condition of step S501 is not satisfied, that is, if the logarithmically converted value lVP does not exceed the median value VMED, the flow shifts to step S503 to determine whether or not the logarithmically converted value lVP is less than the median value VMED. Is done. The determination condition of step S503 is satisfied, that is, the logarithmically converted value lV
If P is smaller than the median value VMED, step S50
4 and the deviation of the median dVMED is updated by the following equation (26).

[0092]

[Expression 26] dVMED ← dVMED-dVm (26)

Step S502 or step S504
After the processing in step, or the determination condition in step S503 is not satisfied, that is, the logarithmically converted value lVP is the median VMED
If it is equal to, the process proceeds to step S505 after step S504 is skipped. In step S505, after the average value VMALL of the logarithmically converted value lVP is updated by the following equation (27), this routine ends. Where N
s is a constant set according to a change in the operating state of the internal combustion engine. For example, if the deviation dlNE of the logarithmic conversion value of the engine speed NE is large, the constant Ns is made small. It is also effective to reduce the constant Ns for a predetermined period after the load or the rate of change of the load becomes a specific condition (for example, a high load or a large rate of change of the load).

[0094]

(27) VMALL ← VMALL + (lVP−VMALL) / Ns (27)

The above processing has the following effects. The knock intensity value is logarithmically converted so that the signal from knock sensor 5 follows a normal distribution, and median value VMED and standard deviation S
Since the knock determination level VJL is created based on the GM, accurate knock determination can be performed without the distribution of the internal combustion engine, cylinder, knock sensor, and the like being affected by changes.
Alternatively, since the median value VMED and the standard deviation SGM are updated for each combustion cycle, the knock determination level V
JL can be set.

By logarithmically converting the knock intensity value by the following equation (28), 10-bit data can be compressed to 8-bit data.

[0097]

[Expression 28] lV = A × log (V / a) (28)

Therefore, it is easy to handle with a 4-bit or 8-bit CPU. Since the operating state of the internal combustion engine is logarithmically converted and the knock determination level is corrected in accordance with the logarithmically converted result, it is possible to prevent a delay in following the knock determination level during a transition. Further, since processing is performed by logarithm, multiplication can be treated as addition and exponentiation can be treated as multiplication, and the load on the CPU can be reduced. The knock determination level VJL can be set to a u value such that the knock determination probability is constant regardless of the operating state of the internal combustion engine in a state where no knock occurs.

As described above, the knock control device for an internal combustion engine according to the present embodiment detects the knock sensor 5 as a signal detecting means for detecting a vibration waveform signal generated in the internal combustion engine 1 and the operating state of the internal combustion engine 1. A crank angle sensor 4b as an operating state detecting means and a vibration waveform signal detected by the knock sensor 5 are input to filter a signal in a frequency band specific to knock, and a peak value V which is a representative value effective for knock detection.
Representative value detection circuit 9 as representative value detection means for detecting P
And a representative value detection circuit 9 based on the operating state of the internal combustion engine 1.
And a knock determination level VJL according to the switching of the filtering frequency band achieved by the ECU 10 that switches the filtering frequency band by the ECU 10.
To change the upper and lower limits of the parameters for setting E
Limit value changing means achieved by the CU 10;
Knock determination means, which is achieved by ECU 10 that determines whether or not knock has occurred in internal combustion engine 1 based on P and knock determination level VJL, and controls the operating state of internal combustion engine 1 according to the determination result by the knock determination means. ECU10
And knock control means achieved by the above. Further, knock determination level VJL is set based on standard deviation SGM as a parameter obtained from peak value VP. The upper limit value SGMmax and the lower limit value SGMmin of the standard deviation SGM are set for each center frequency f0 of the frequency band unique to knock.

That is, a signal in a frequency band peculiar to knock is filtered by a representative value detection circuit 9 from a vibration waveform signal generated in the internal combustion engine 1 detected by the knock sensor 5, and a peak value VP effective for knock detection is detected. Is done. Here, based on the operation state of the internal combustion engine 1, the representative value detection circuit 9
The upper limit value SGMmax and the lower limit value SGMmin are changed so that the standard deviation SGM as a parameter obtained from the peak value VP is set within the optimum range. By comparing the knock determination level VJL thus set with the peak value VP, the presence / absence of knock in the internal combustion engine 1 is accurately determined. According to the determination result thus obtained, the operating state of the internal combustion engine 1 can be appropriately controlled.

[Brief description of the drawings]

FIG. 1 is a schematic diagram showing a configuration of an internal combustion engine to which an internal combustion engine knock control device according to one embodiment of the present invention is applied.

FIG. 2 is a block diagram illustrating a configuration of a representative value detection circuit in FIG. 1;

FIG. 3 is a flowchart showing a processing procedure of knock detection / judgment control by a CPU in an ECU used in a knock control device for an internal combustion engine according to an embodiment of the present invention.

FIG. 4 is a characteristic diagram showing a lognormal distribution of peak values in the logarithmic conversion processing of FIG. 3;

FIG. 5 is a characteristic diagram showing a normal distribution obtained by converting the lognormal distribution of FIG.

FIG. 6 is a flowchart showing a processing procedure for setting a knock determination level in FIG. 3;

FIG. 7 is a characteristic diagram showing a relationship between an engine speed and a u value in a knock determination level setting process of FIG. 6 using a knock detection cycle as a parameter.

FIG. 8 is a flowchart showing a processing procedure of knock determination in FIG.

FIG. 9 is a characteristic diagram showing a frequency distribution of logarithmically converted values in the knock determination processing of FIG.

FIG. 10 is a flowchart showing a processing procedure for updating a standard deviation in FIG. 3;

FIG. 11 is an explanatory diagram showing that a standard deviation is obtained by the processing of FIG. 10;

FIG. 12 is a flowchart showing a processing procedure for updating the median value in FIG. 3;

[Explanation of symbols]

 Reference Signs List 1 internal combustion engine 4b crank angle sensor (operating state detecting means) 5 knock sensor (signal detecting means) 9 representative value detecting circuit (representative value detecting means) 10 ECU (electronic control unit) 11 CPU

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) F02P 17/12 (72) Inventor Hideki Yukimoto 1-1-1, Showa-cho, Kariya-shi, Aichi Prefecture Inside DENSO Corporation (72) Inventor Naoki Kokubo 1-1-1, Showa-cho, Kariya-shi, Aichi Prefecture Inside DENSO Corporation (72) Inventor Kenji Kasajima 1-Toyota-cho, Toyota-shi, Aichi Prefecture Toyota Motor Corporation F-term (reference) 3G019 GA14 3G022 DA02 EA02 FA01 GA13 3G084 BA17 DA38 EA11 EC04 FA07 FA20 FA25 FA33 FA38

Claims (3)

[Claims] 1. A signal detection means for detecting a vibration waveform signal generated in an internal combustion engine; an operation state detection means for detecting an operation state of the internal combustion engine; and an input of the vibration waveform signal detected by the signal detection means. A representative value detecting means for filtering a signal in a frequency band peculiar to knock and detecting a representative value effective for knock detection; and switching a filtering frequency band in the representative value detecting means based on an operation state of the internal combustion engine. Frequency switching means, Limit value changing means for changing upper and lower limits of a parameter for setting a knock determination level in accordance with switching of the filtering frequency band by the frequency switching means, The representative value and the knock determination level Knock determination means for determining whether or not knock has occurred in the internal combustion engine based on A knock control device for an internal combustion engine, comprising: knock control means for controlling an operation state of a fuel engine. 2. The knock control device for an internal combustion engine according to claim 1, wherein the knock determination level is set based on a standard deviation value as the parameter obtained from the representative value. 3. The knock control device for an internal combustion engine according to claim 2, wherein the standard deviation value has upper and lower limits set for each center frequency of a frequency band specific to knock.