JP2003097343A - Misfire detecting device for internal combustion engine - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve accuracy for misfire detection. SOLUTION: Ion current generated by combustion in an engine cylinder is detected by an ignition plug. An actual frequency [a(%)] where an ion current-peak value [Pi] is equal or below a misfire criterion value [Vth] is obtained, and linear distribution characteristics of the ion current-peak value [Pi] at the time of normal combustion are predicted from a frequency 50% point [v50] and a standard deviation value [σ] to obtain a predicted frequency [A(%)] where the linear distribution of the ion current-peak value [Pi] is equal or below the misfire criterion value [Vth]. The predicted frequency [A(%)] is equivalent to a cumulative frequency of the combustion ion that is equal or below the misfire criterion value [Vth]. The difference [Δ(%)] between the actual frequency [a(%)] and the predicted frequency [A(%)] is obtained to determine any misfire based on whether or not the difference [Δ(%)] is greater than the criterion value. The difference [Δ(%)] is equivalent to a cumulative frequency of misfire alone where the cumulative frequency of the combustion ion as an error is subtracted from the actual frequency [a(%)] in which the peak value [Pi] is equal or below the misfire criterion value [Vth].

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an internal combustion engine misfire detection device for detecting an internal combustion engine misfire.

[0002]

2. Description of the Related Art In recent years, focusing on the characteristic that ions are generated when a mixture is burned in a cylinder of an internal combustion engine, the ionic current generated in the cylinder at each ignition is detected through an electrode of a spark plug. , A technique for detecting ignition / misfire based on the detected ion current value has been developed. The conventional ignition / misfire determination method utilizes the property that the ion current increases at ignition and the ion current decreases at the time of misfire, and the detected ion current peak value is compared with a predetermined determination value to determine the ion current peak. If the value is greater than or equal to the determination value, it is determined that ignition has occurred, and if not, it is determined that misfire has occurred.

[0003]

However, when a deposit adheres to the electrode of the spark plug or when the temperature is high and the humidity is high, the ion current detected through the electrode of the spark plug decreases, so the peak value of the detected ion current is reduced. In the conventional misfire detection method that compares with a predetermined determination value, there is a possibility that misfire is erroneously detected even when a deposit is attached to the spark plug, high temperature and high humidity, and even in a normal combustion state. Further, there is a possibility that the ion current peak value may change and misfire may be erroneously detected due to variations in manufacturing of the ion current detection system or changes over time.

The present invention has been made in view of the above circumstances, and therefore an object thereof is to provide a misfire detecting device for an internal combustion engine which can improve the accuracy of misfire detection.

[0005]

In order to achieve the above object, an internal combustion engine misfire detection device according to claim 1 of the present invention detects the combustion state of the internal combustion engine by a combustion state detection means and the combustion state. The combustion state detection value for each combustion cycle detected by the detection means is statistically processed by the distribution characteristic evaluation means to evaluate the distribution characteristic of the combustion state detection value, and the misfire determination means ensures that the distribution characteristic of the combustion state detection value is normal. The presence / absence of misfire is determined by determining the distribution characteristic at the time of combustion or the distribution characteristic at the time of occurrence of misfire. In other words, even if the output of the combustion state detection means (combustion state detection value) is relatively decreased due to manufacturing variation of the combustion state detection means, change over time, temperature, humidity, etc., the distribution characteristic of the combustion state detection value shows normal combustion. Since the distribution characteristics differ depending on the time and the occurrence of misfire, if the presence or absence of misfire is judged from the distribution characteristics of the combustion state detection value,
The misfire can be accurately detected without being affected by manufacturing variations of the combustion state detecting means, changes with time, temperature, humidity and the like.

In this case, the combustion state detecting means detects the ion current generated by the combustion in the cylinder and uses the ion current peak value as the combustion state detection value. The logarithmic normal distribution statistically processed value may be obtained as the distribution characteristic value of the combustion state detection value. That is, since the distribution characteristic of the ion current peak value at the time of normal combustion has an exponential function characteristic, if the statistically processed value of the lognormal distribution of this ion current peak value is obtained, it is possible to accurately determine whether it is normal combustion or misfire. You can

Further, as in claim 3, a frequency 50% point [v50] and a standard deviation value [σ] in the distribution characteristic of the combustion state detection value are calculated, and these frequency 50% point [v50] and the standard deviation are calculated. The distribution characteristic of the combustion state detection value at the time of normal combustion is predicted from the value [σ], and the combustion state detection value is the misfire determination value [Vt
h] and the predicted frequency [A (%)] that is less than
The actual frequency [a (%)] at which the actual combustion state detection value is equal to or less than the misfire determination value [Vth] is calculated, and the difference [Δ] between the actual frequency [a (%)] and the predicted frequency [A (%)]. (%) = A (%)-
A (%)] may be obtained and the presence or absence of misfire may be determined based on this difference [Δ (%)]. The difference [Δ (%)] thus obtained corresponds to the frequency increase due to the misfire, and therefore the misfire can be accurately detected from the difference [Δ (%)].

Further, as in claim 4, the difference [Δ
(%)], The misfire occurrence rate may be calculated. In other words, as the number of misfires increases, the difference [Δ
(%)] Becomes large, the misfire occurrence rate can be calculated accurately based on the difference [Δ (%)], and appropriate measures can be taken according to the misfire occurrence rate.

According to a fifth aspect, the difference [Δ
(%)] For determining the presence or absence of misfire, and means for determining the presence or absence of misfire based on the actual frequency [a (%)] or the combustion state detection value. The switching may be performed according to the distribution characteristic value such as the% point [v50]. That is, the frequency 50% point [v
As the difference between 50] and the misfire determination value [Vth] increases, the noise component included in the actual frequency [a (%)] decreases, so that the frequency 50% point [v50] and the misfire determination value [Vth] In a region where the difference is large, the presence or absence of misfire can be accurately determined based on the actual frequency [a (%)] or the combustion state detection value without using the distribution characteristic of the combustion state detection value.

Further, as in claim 6, the frequency is 50%.
Whether or not there is a misfire based on the difference [Δ (%)] when a stable distribution characteristic of the combustion state detection value is obtained in which the point [v50] and the standard deviation value [σ] are each within a predetermined range. May be determined. If you do this,
The accuracy of misfire determination can be further improved.

Further, as described in claim 7, the drift value (zero point error) of the output of the combustion state detecting means is detected by the drift value detecting means, and the output of the combustion state detecting means is detected by the output correcting means. It is also possible to calculate the combustion state detection value by correcting the distribution state of the combustion state detection value by statistically processing the combustion state detection value for each combustion cycle corrected by the output correcting means. Thus, if the output of the combustion state detection means is corrected by the drift value to obtain the combustion state detection value, a net combustion state detection value without drift (zero point error) can be obtained, and the misfire determination accuracy can be further improved. Can be improved.

Further, as described in claim 8, a frequency 50 in the distribution characteristic of the drift value detected by the drift value detecting means.
When the% point [c50] and the standard deviation value [σc] are calculated, and the frequency 50% point [c50] and the standard deviation value [σc] are not within the predetermined ranges, respectively, the drift value distribution characteristics become unsatisfactory. It may be judged as stable and the misfire determination by the misfire determination means may be prohibited by the misfire determination prohibiting means. That is,
When the drift value distribution characteristic becomes unstable, the correction accuracy of the combustion state detection value based on the drift value deteriorates, and the misfire determination accuracy decreases. Therefore, if the distribution characteristic of the drift value is judged to be unstable from the frequency 50% point [c50] and the standard deviation value [σc] in the distribution characteristic of the drift value, the misfire determination is prohibited to prevent the misfire. Detection can be prevented in advance.

Further, as described in claim 9, the output (ion current detection value) of the combustion state detecting means during the intake stroke or the compression stroke may be detected as a drift value. In other words, since combustion ions are not generated during the intake stroke or compression stroke, it is considered that the current detected during the intake stroke or compression stroke is not due to combustion ions but due to noise components such as smoldering leakage current. To be Therefore, if the output (ion current detection value) of the combustion state detecting means during the intake stroke or the compression stroke is detected as the drift value, the drift value can be accurately detected.

According to a tenth aspect of the present invention, the in-cylinder pressure is detected at a plurality of timings, and based on the in-cylinder pressure detection values, any one of the work rate, the heat generation rate, the indicated mean effective pressure, and the work amount. May be calculated and used as the combustion state detection value. With this configuration, it is possible to accurately determine the presence or absence of misfire from the detected value of the in-cylinder pressure by the same method as in the case of detecting the ion current.

In this case, as in the eleventh aspect, at least one of offset correction, gain correction and hysteresis correction may be performed on the in-cylinder pressure detection value. With this configuration, the accuracy of the combustion state detection value obtained from the in-cylinder pressure detection value can be increased, and the misfire determination accuracy can be further improved.

The misfire determination value [Vth] may be a fixed value set in advance for the sake of simplifying the calculation process, but the combustion state detection value detected by the combustion state detection means is the elapsed time of the combustion state detection means. It may shift due to changes (existence of smoldering, etc.) and variations in combustion conditions (Fig. 2
0), if the misfire determination value [Vth] is a fixed value,
When the deviation of the combustion state detection value due to the change with time of the combustion state detecting means or the variation of the combustion state becomes large, there is a concern that the misfire cycle may be erroneously determined as the normal combustion cycle.

Therefore, as in claim 12, the misfire determination value [Vth] may be calculated by the misfire determination value calculation means based on the distribution characteristic of the combustion state detection values evaluated by the distribution characteristic evaluation means. . By doing so, even if the distribution characteristics of the combustion state detection values are deviated due to changes over time in the combustion state detection means, variations in the combustion state, etc., the misfire determination value [Vth] can be set in consideration of the deviation. The misfire cycle can be accurately discriminated from the normal combustion cycle without being affected by the deviation of the combustion state detection value due to the aging of the combustion state detecting means or the variation of the combustion state, etc. It is possible to make a misfire determination with high property.

In this case, the distribution characteristic pattern of the combustion state detection value evaluated by the distribution characteristic evaluation means is separated into a normal combustion cycle distribution pattern and a misfire cycle distribution pattern (see FIG. 20). It is advisable to calculate the misfire determination value [Vth] based on the misfire cycle distribution pattern. By doing so, an appropriate misfire determination value [Vth] can be set in accordance with the distribution pattern of the actual misfire cycle, and the misfire cycle can be accurately detected.

A specific method of setting the misfire determination value [Vth] is, as in claim 14, the maximum combustion state detection value among the combustion state detection values belonging to the distribution pattern of the misfire cycle or a value slightly larger than that. Is a misfire determination value [Vth]. As a result, it is possible to determine all combustion state detection values when a misfire actually occurs as a misfire cycle, and also to prevent a normal combustion cycle and a noise component from being erroneously determined as a misfire cycle.

By the way, the combustion state detection value may fluctuate depending on the operating state of the internal combustion engine. Especially, during unstable combustion, the combustion state detection value tends to have a large variation, but the combustion state detection value varies. When becomes larger, the width (spread) of both distribution patterns of the normal combustion cycle and the misfire cycle becomes larger, and the intervals of both distribution patterns come closer to each other. As a result, if the distance between the two distribution patterns becomes too narrow, or if the two distribution patterns overlap as shown in FIG. 22, it becomes difficult to accurately distinguish between the normal combustion cycle and the misfire cycle.

Therefore, according to the fifteenth aspect, the calculation of the misfire determination value [Vth] is permitted or prohibited based on the operating state of the internal combustion engine and / or the distribution characteristic of the combustion state detection value evaluated by the distribution characteristic evaluation means. The permission / prohibition determination means may determine whether to perform. With this configuration, the misfire determination value [Vth] is calculated only when the variation in the combustion state detection value is small, and the misfire determination value [Vth] is not calculated when the variation in the combustion state detection value is large. It is possible to prevent a decrease in misfire determination accuracy due to variations in the combustion state detection values.

Specifically, when the degree of separation between the distribution pattern of the normal combustion cycle and the distribution pattern of the misfire cycle is equal to or more than a predetermined value, the misfire determination value [Vth].
It is advisable to permit the calculation of In this way, the misfire determination value [Vth] can be calculated only when the distribution patterns of the normal combustion cycle and the misfire cycle are clearly separated, and the accurate misfire determination value [Vth] is obtained. be able to.

In this case, the degree of separation between the normal combustion cycle distribution pattern and the misfire cycle distribution pattern is determined based on the width of the noise region between both distribution patterns and the height of both distribution patterns. It may be judged. By doing so, the degree of separation between both distribution patterns can be accurately determined.

According to the eighteenth aspect of the present invention, a plurality of misfire determination values [Vth] including the latest misfire determination value [Vth] calculated by the misfire determination value calculation means are stored in the storage means and stored in the storage means. The misfire determination value [Vth] to be used this time from the plurality of misfire determination values [Vth] based on the distribution characteristic of the combustion state detection value evaluated by the operating state and / or distribution characteristic evaluation means.
May be selected by the misfire determination value selection means. With this configuration, the optimum misfire determination value [Vth] can be selected from the plurality of misfire determination values [Vth] according to the variation of the combustion state detected value, and the variation of the combustion state detected value, etc. It is possible to make a highly reliable misfire determination that is not easily affected by.

In this case, when the drift value of the output of the combustion state detecting means is smaller than a predetermined value, the initial set value is selected from the plurality of misfire determination values [Vth]. You can As the combustion state detection means, for example, when detecting an ion current using a spark plug, if smoldering of the spark plug occurs, the drift value of the ion current detection value that is the combustion state detection value increases, and smolder When it does not occur, the drift value of the ion current detection value tends to be small. As shown in FIG. 20 (a), when smoldering does not occur, the distribution pattern of the misfire cycle falls below the initial set value for misfire determination, so the drift value of the output of the combustion state detecting means (for example, the degree of smoldering). If the initial value is selected from a plurality of misfire determination values [Vth] when is smaller than a predetermined value, the misfire cycle can be accurately detected when smoldering or the like is not occurring. In addition to being able to do, the appropriate misfire judgment value [Vt
h] can be selected, and misfire can be appropriately determined even when smoldering occurs.

At this time, when the drift value of the output of the combustion state detecting means is equal to or more than a predetermined value, the latest misfire determination value [Vth] calculated by the misfire determination value calculating means is selected. You can do it. In this way, when smoldering occurs, the latest misfire determination value [Vth] corresponding to the degree of smoldering can be selected, and accurate misfire determination can be performed even when smoldering occurs.

In the inventions according to claims 12 to 20 described above, the misfire determination value calculating means for calculating the misfire determination value [Vth] based on the distribution characteristic of the combustion state detection value is provided. A plurality of misfire determination values [Vth] are prepared in advance without providing any means, and as shown in claim 21, the operating state of the internal combustion engine and / or the combustion state detection value evaluated by the distribution characteristic evaluation means The misfire determination value [Vth] to be used this time may be selected by the misfire determination value selection means from the plurality of misfire determination values [Vth] based on the distribution characteristics. Even in this case, the optimum misfire determination value [Vth] can be selected from a plurality of misfire determination values [Vth] according to the variation in the combustion state detected value, and the variation in the combustion state detected value, etc. It is possible to make a highly reliable misfire determination with little influence of.

[0028]

DESCRIPTION OF THE PREFERRED EMBODIMENTS << Embodiment (1) >> An embodiment (1) of the present invention will be described below with reference to FIGS.
First, the circuit configuration of the ignition control system will be described with reference to FIG. One end of the primary coil 22 of the ignition coil 21 is connected to the battery 23, and the other end of the primary coil 22 is connected to the collector of the power transistor 25 incorporated in the igniter 24. One end of the secondary coil 26 is connected to the spark plug 27, and the other end of the secondary coil 26 is connected to the ground via the two Zener diodes 28 and 29.

The two Zener diodes 28 and 29 are connected in series in the opposite directions to each other, the capacitor 30 is connected in parallel to one Zener diode 28, and the ion current detection resistor 31 is connected in parallel to the other Zener diode 29. There is. Capacitor 30 and ion current detection resistor 3
The potential Vin between 1 and the inverting amplifier circuit 3 via the resistor 32
3 is input to the inverting input terminal (-) of the inverter 3 and is inverted and amplified. The output voltage V of the inverting amplifier circuit 33 is input to the engine control circuit 34 as an ion current detection signal. The ion current detection circuit 35 includes zener diodes 28 and 29, a capacitor 30, an ion current detection resistor 31, and an inverting amplifier circuit 3.
3 and the like, and serves as a combustion state detecting means for detecting the combustion state by the ion current detection signal.

During engine operation, the engine control circuit 34
The power transistor 25 is turned on / off at the rising / falling edge of the ignition command signal transmitted from the igniter 24 to the igniter 24. When the power transistor 25 is turned on, a primary current flows from the battery 23 to the primary coil 22, and then,
When the power transistor 25 turns off, the primary coil 22
The primary current is cut off, a high voltage is electromagnetically induced in the secondary coil 26, and the high voltage causes a spark discharge between the electrodes 36 and 37 of the spark plug 27. The spark discharge current flows from the ground electrode 37 of the spark plug 27 to the center electrode 36, charges the capacitor 30 via the secondary coil 26, and flows to the ground side via the Zener diodes 28 and 29. After charging the capacitor 30, the ion current detection circuit 3 uses the charging voltage of the capacitor 30 regulated by the Zener voltage of the Zener diode 28 as a power source.
5 is driven, and the ion current is detected as described later.

On the other hand, the ionic current flows in the direction opposite to the spark discharge current. That is, after ignition is completed, the electrode 3 of the spark plug 27 is charged by the charging voltage of the capacitor 30.
Since a voltage is applied between 6 and 37, an ion current flows between the electrodes 36 and 37 due to the ions generated when the air-fuel mixture burns in the cylinder. This ion current flows from the center electrode 36 to the ground electrode 37. To the capacitor 30 through the ion current detection resistor 31 from the ground side. At this time, the input potential Vin of the inverting amplification circuit 33 changes according to the change of the ionic current flowing through the ionic current detection resistor 31, and the voltage corresponding to the ionic current is output from the output terminal of the inverting amplification circuit 33 to the engine control circuit 34. To be done.

A noise mask 38, a peak hold circuit 39, an A / D converter 40 and a microcomputer 41 are built in the engine control circuit 34. The output voltage of the ion current detection circuit 35 is input to the peak hold circuit 39 after the noise component is removed by the noise mask 38. The peak hold circuit 39 has a peak value of the output voltage of the noise mask 38 (ion current peak value).
It detects Pi and holds it (see FIG. 2). The output of the peak hold circuit 39 is read into the microcomputer 41 via the A / D converter 40.

The ROM (storage medium) of the microcomputer 41 stores various engine control routines for performing fuel injection control and ignition timing control.
A misfire determination routine shown in FIGS. 5 and 6 described later is stored. By executing this misfire determination routine by the microcomputer 41, it is determined whether or not there is a misfire,
When it is finally determined that there is a misfire, the warning lamp 42 is turned on (or blinks) to warn the driver.

Next, the method of misfire determination of this embodiment (1) will be described. In the present embodiment (1), the ion current peak value [Pi] peak-held by the peak hold circuit 39 of the ion current detection circuit 35 is used as the combustion state detection value, and as shown in FIG. [P
i] is used as the distribution characteristic value of the combustion state detection value. Since the distribution characteristic of the ion current peak value [Pi] at the time of normal combustion has an exponential function characteristic, if the logarithmic normal distribution of this ion current peak value [Pi] is statistically processed to obtain the cumulative frequency, As shown in FIGS. 3 and 4, the cumulative frequency has a linear distribution. Then, in order to evaluate the characteristics of the lognormal distribution of the ion current peak value [Pi], the frequency 50% point [v
50] and the standard deviation value [σ] are calculated.

On the other hand, as shown in (4) of FIG. 4, when a misfire occurs, there are two peaks in the distribution of the ion current peak value [Pi] to be detected, and combustion ions (ion current peak value [Pi] due to combustion) In addition to the peaks of the distribution of (1), peaks of the distribution of the ion current peak value [Pi] are generated due to the misfire. Normally, the peaks of the distribution of combustion ions occur in the region of the misfire determination value [Vth] or more, and the peaks of the distribution of misfire are the misfire determination value [Vt].
h] occurs in the following areas. As a result, when a misfire occurs, the cumulative frequency of the ion current peak value [Pi] does not have a linear distribution, and the cumulative frequency of the region below the misfire determination value [Vth] increases. Generally, the misfire determination value [Vth] is set near the maximum value of the ion current peak value [Pi] at the time of occurrence of misfire or a value slightly larger than that.

As shown in FIG. 4, when the distribution of combustion ions is completely separated from the distribution of misfires, the cumulative frequency of misfires can be calculated accurately without being affected by the combustion ions. As shown in FIG. 4, the spark plug 2
When the deposit of 7 and the environment of high temperature and high humidity, etc., the combustion ions generally decrease and the peak of the distribution of combustion ions approaches the misfire determination value [Vth], a part of the peak of the distribution of combustion ions is misfired. Misfire judgment value [Vth] overlapping with distribution peaks
The cumulative frequency of the following areas is higher than the cumulative frequency of actual misfires.

Therefore, as shown in FIG. 4, the actual frequency [a (%)] at which the actual ion current peak value [Pi] is less than or equal to the misfire determination value [Vth] is calculated, and the frequency is 50%.
The linear distribution characteristic of the ion current peak value [Pi] during normal combustion is predicted from the point [v50] and the standard deviation value [σ], and the linear distribution of the ion current peak value [Pi] is misfired. A prediction frequency [A (%)] that is equal to or lower than the determination value [Vth] is calculated. The predicted frequency [A (%)] corresponds to the cumulative frequency of the combustion ions that is equal to or less than the misfire determination value [Vth].
Then, the actual frequency [a (%)] and the predicted frequency [A
(%)] And the difference [Δ (%)]. Δ (%) = a (%) − A (%)

This difference [Δ (%)] is the misfire determination value [V
th] or less and the actual frequency [a (%)] is equivalent to the cumulative frequency of misfire only, which is obtained by removing the cumulative frequency of combustion ions, which is an error. As shown in FIG. 4, when the distribution of burning ions is completely separated from the distribution of misfire, the predicted frequency [A (%)] is 0 (%), so Δ (%) = a
(%).

As shown in FIG. 4, when the combustion ions are large and the distribution of the combustion ions is completely separated from the distribution of the misfire, there is no false detection of the misfire, so the peak of the ion current detection circuit 35. It is general to saturate the ion current peak value [Pi] detected by the hold circuit 39. In the region where the ion current peak value [Pi] saturates, the standard deviation value [σ] becomes 0, so the frequency 50% point [v5] of the ion current peak value [Pi]
0] is equal to or higher than the saturating point PM, the difference [Δ
(%)] Based on the actual frequency [a (%)] at which the actual ion current peak value [Pi] is less than or equal to the misfire determination value [Vth].

The above-described misfire determination is executed by the misfire determination routine shown in FIGS. 5 and 6. This misfire determination routine is executed for each combustion cycle. First, in step 101, the ion current peak value [Pi] output from the peak hold circuit 39 of the ion current detection circuit 35 is read as a combustion state detection value, and the next step 102
Then, it is determined whether or not the calculation of the distribution of the ion current peak value [Pi] is permitted. If calculation of the distribution of the ion current peak value [Pi] is prohibited, such as when the fuel is cut, the process proceeds to step 104, and the Pi table that stores the data of the ion current peak value [Pi] up to the previous time is initialized. Then, this routine is finished.

On the other hand, if the calculation of the distribution of the ion current peak value [Pi] is permitted, the routine proceeds to step 103, where P
The data of the current peak value [Pi] of the ion current is stored in the i table.

After that, the routine proceeds to step 105, where it is judged whether or not the number of pieces of data of the ion current peak value [Pi] accumulated in the Pi table has become a predetermined number (eg, 1000) or more, and the number of data is made a predetermined number or more. If not, the present routine is terminated without performing the subsequent processing. Then, the ion current peak value [P
When the number of data of [i] exceeds a predetermined number, the routine proceeds to step 106, where the frequency 50% point [v50] and the standard deviation value [σ] in the distribution characteristic of the ion current peak value [Pi] are calculated, In the next step 107, the distribution characteristic of the ion current peak value [Pi] during normal combustion is predicted from the frequency 50% point [v50] and the standard deviation value [σ] to determine the ion current peak value [Pi]. The predicted frequency [A (%)] that is less than or equal to the misfire determination value [Vth] is calculated, and the actual frequency [a (%)] at which the actual ion current peak value [Pi] is less than or equal to the misfire determination value [Vth] is calculated. Calculated and this actual frequency [a (%)]
[Δ (%) = a (%)] between the prediction frequency and the prediction frequency [A (%)]
-A (%)] is calculated. These steps 106, 10
The processing of 7 serves as a distribution characteristic evaluation means in the claims.

After that, the routine proceeds to step 108 of FIG. 6, and it is judged whether or not the misfire detection condition is satisfied, for example, by whether or not the cooling water temperature is equal to or higher than a predetermined temperature, and if the misfire detection condition is not satisfied. , This routine ends without performing the subsequent processing.

On the other hand, if the misfire detection condition is satisfied, the routine proceeds to step 109, where it is judged whether or not the misfire judgment execution condition based on the distribution characteristic of the ion current peak value [Pi] is satisfied. Here, as the misfire determination execution condition based on the distribution characteristic of the ion current peak value [Pi], for example,
(1) As shown in FIG. 4, the frequency 50% point [v50] is smaller than the saturating point PM, and (2) it is during idle operation. Either of these two conditions (1) and (2) On the other hand, if the conditions are satisfied, the ion current peak value [P
i]], the misfire determination execution condition based on the distribution characteristics of
If both conditions (1) and (2) are not satisfied, the misfire determination execution condition based on the distribution characteristic of the ion current peak value [Pi] is not satisfied. Since the ion current peak value [Pi] becomes relatively small during idle operation, accurate misfire determination is performed by performing misfire determination based on the distribution characteristics.

If it is determined in step 109 that the misfire determination execution condition based on the distribution characteristic of the ion current peak value [Pi] is satisfied, the process proceeds to step 110 and the actual frequency [a (%)] And the prediction frequency [A (%)], the difference [Δ (%)] is compared with the first misfire determination value [C1]. This first misfire determination value [C1]
Is set to the minimum misfire frequency (for example, 3%) that may worsen emissions. If the difference [Δ
(%)] Is larger than the first misfire determination value [C1], it is determined that a misfire has occurred, and the routine proceeds to step 111, where the misfire determination flag MFT is set to "1" which means a misfire has occurred.

On the other hand, if the difference [Δ (%)] is less than the first misfire determination value [C1], there is no misfire (normal combustion).
Therefore, the process proceeds to step 112, and the misfire determination flag M
Set FT to "0", which means no misfire. The processing of these steps 110 to 112 serves as a misfire determination means in the claims.

After this, the routine proceeds to step 116, where the difference [Δ
(%)] Is larger than the third misfire determination value [C3]. The third misfire determination value [C3] is set to the minimum misfire frequency (for example, 25%) that may cause catalyst melting loss. If the difference [Δ (%)] is larger than the third misfire determination value [C3], the routine proceeds to step 117, where the fuel cut flag MFC is set to "1" and the fuel cut is executed. On the other hand, if the difference [Δ (%)] is less than or equal to the third misfire determination value [C3], step 118
And set the fuel cut flag MFC to "0",
Do not execute fuel cut.

On the other hand, if it is determined in step 109 that the misfire determination execution condition based on the distribution characteristic of the ion current peak value [Pi] is not satisfied, the process proceeds to step 113 and the actual frequency [a (%)]. Is compared with the second misfire determination value [C2]. The second misfire determination value [C2] is set to a value substantially the same as the first misfire determination value [C1] or a value slightly smaller than that. If the actual frequency [a (%)] is larger than the second misfire determination value [C2], it is determined that a misfire has occurred and the routine proceeds to step 114, where the misfire determination flag M
Set FT to "1". On the other hand, the actual frequency [a
(%)] Is equal to or less than the second misfire determination value [C2], it is determined that there is no misfire, the process proceeds to step 115, and the misfire determination flag MFT is set to "0".

After that, the routine proceeds to step 119, where the actual frequency [a (%)] is compared with the third misfire determination value [C3], and the actual frequency [a (%)] is the third misfire determination value [C3]. ], The fuel cut flag M
FC is set to "1" and fuel cut is executed. On the other hand, the actual frequency [a (%)] is the third misfire determination value [C
3] If it is equal to or less than that, the process proceeds to step 121, the fuel cut flag MFC is set to "0", and the fuel cut is not executed.

According to the embodiment (1) described above,
While predicting the distribution characteristics of the ion current peak value [Pi] during normal combustion, the predicted frequency [A (%)] at which the ion current peak value [Pi] becomes equal to or less than the misfire determination value [Vth] is calculated, and The actual frequency [a (%)] at which the ionic current peak value [Pi] of is less than or equal to the misfire determination value [Vth] is calculated, and the actual frequency [a (%)] and the predicted frequency [A (%)] are calculated. The difference [Δ (%) = a (%) − A (%)] is calculated. Since this difference [Δ (%)] corresponds to the actual misfire frequency obtained by removing the cumulative frequency of combustion ions, which is an error, from the actual frequency [a (%)] that is less than or equal to the misfire determination value [Vth]. , In the region where the actual frequency [a (%)] is larger than the actual misfire frequency (the region where the misfire determination execution condition based on the distribution characteristic is satisfied), the difference [Δ (%)] is the first misfire determination value [ C1
], It is determined whether or not there is a misfire that may worsen the emission. As a result, as shown in FIG. 4, the actual frequency [a] at which the actual ion current peak value [Pi] becomes equal to or less than the misfire determination value [Vth].
(%)] Can also be used to determine the misfire based on the difference [Δ (%)], which is the actual misfire frequency obtained by removing the error, even in the area where the cumulative frequency of combustion ions, which is the error, is included. It is possible to accurately detect misfire.

In the present embodiment (1), as shown in FIG. 4, the actual frequency [a (%)] is substantially the same as the actual misfire frequency (difference [Δ (%)]) ( In the region where the condition for executing the misfire determination based on the distribution characteristic is not satisfied), the actual frequency [a (%)] is compared with the second misfire determination value [C2] to perform the misfire determination. Value [Pi
Whether or not there is a misfire may be determined by whether or not] is less than or equal to the misfire determination value [Vth].

The actual frequency [a (%)] and the predicted frequency [A
(%)], The misfire occurrence rate (misfire frequency) may be calculated based on the difference [Δ (%)].

<Embodiment (2)> In the above embodiment (1), whether the frequency 50% point [v50] is smaller than the saturating point PM (whether or not the misfire determination execution condition based on the distribution characteristic is satisfied). However, in the embodiment (2) of the present invention, as shown in the misfire determination method switching map of FIG. 7, the frequency 50% point [v50].
According to the standard deviation value [σ], three regions I, II, and III are divided, and different misfire determination methods are used in each region I, II, and III.

In the misfire determination method switching map of FIG. 7, the region I is a region where the frequency 50% point [v50] is larger than the saturating point PM regardless of the size of the standard deviation value [σ]. In the region II, the frequency 50% point [v50] is in the range from the point PL (see FIG. 4) at which the misfire determination value [Vth] is slightly larger to the saturating point PM, and the standard deviation value [σ] is predetermined. This is a region where stable distribution characteristics below the range can be obtained. On the other hand, the region III is the above two regions I,
This is the area excluding II. Therefore, the area below the point PL is the area III regardless of the size of the standard deviation value [σ].
In the range of PL <frequency 50% point [v50] <PM, the area where the standard deviation value [σ] is equal to or smaller than the predetermined range is the area II, and the area where the standard deviation value [σ] is equal to or larger than the predetermined range is the area III.

In this case, in region I, the presence or absence of misfire is determined by whether or not the ion current peak value [Pi] is less than or equal to the misfire determination value [Vth]. In the area II, the actual frequency [a
(%)] And the prediction frequency [A (%)] difference [Δ (%)]
Is larger than the first misfire determination value [C1], it is determined whether or not a misfire that may worsen the emission has occurred. Also, as the frequency of misfires increases, the frequency becomes 5
Since the 0% point [v50] becomes small and the standard deviation value [σ] becomes large, it is judged that there is a complete misfire in region III, and a
(%) = 100% is set. This region III includes a region on the way to complete misfire, but it is determined to be complete misfire early in order to prevent catalyst melting loss.

In this embodiment (2), after executing the processing of steps 101 to 107 of the misfire determination routine of FIG. 5, the routine proceeds to step 130 of FIG. 8 to check whether the misfire detection condition is satisfied. For example, it is determined whether or not the cooling water temperature is equal to or higher than a predetermined temperature, and if the misfire detection condition is not satisfied, this routine is terminated without performing the subsequent processing.

On the other hand, if the misfire detection condition is satisfied, the routine proceeds to step 131, where the misfire of FIG. 7 is performed based on the frequency 50% point [v50] of the ion current peak value [Pi] and the standard deviation value [σ]. It is determined which one of the three areas I, II and III divided by the determination method switching map belongs to.

As a result, if the region I is determined, the routine proceeds to step 132, where the ion current peak value [Pi] is compared with the misfire determination value [Vth], and the ion current peak value [Pi] is determined as the misfire determination value [Vth. ], It is determined that a misfire has occurred, the process proceeds to step 133, and the misfire determination flag MFT is set to "1". On the other hand, if the ion current peak value [Pi] is greater than or equal to the misfire determination value [Vth], it is determined that there is no misfire, and the routine proceeds to step 134, where the misfire determination flag M
Set FT to "0".

Thereafter, the process proceeds to step 135, and the process shown in FIG.
The actual frequency [a (%)] calculated in step 107
If the actual frequency [a (%)] is larger than the third misfire determination value [C3], the routine proceeds to step 136, where the fuel cut flag MFC is set to "1". To cut fuel. On the other hand, the actual frequency [a
(%)] Is less than or equal to the third misfire determination value [C3], the routine proceeds to step 137, where the fuel cut flag MFC is set to "0".
Set to and do not execute fuel cut.

On the other hand, if it is determined to be the region II in the above step 131, the process proceeds to step 138, and the actual frequency [a (%)] calculated in step 107 of FIG. 5 and the predicted frequency [A
(%)] And the difference [Δ (%)] from the first misfire determination value [C
1] and if the difference [Δ (%)] is larger than the first misfire determination value [C1], it is determined that a misfire has occurred, the process proceeds to step 139, and the misfire determination flag MFT is set to "1". To do. On the other hand, if the difference [Δ (%)] is less than or equal to the first misfire determination value [C1], it is determined that there is no misfire, the process proceeds to step 140, and the misfire determination flag MFT is set to "0".

Thereafter, the process proceeds to step 141, where the difference [Δ
(%)] Is compared with the third misfire determination value [C3]. If the difference [Δ (%)] is larger than the third misfire determination value [C3], the routine proceeds to step 142, where the fuel cut flag MFC
Is set to "1" to execute fuel cut. On the other hand, if the difference [Δ (%)] is less than or equal to the third misfire determination value [C3], the routine proceeds to step 143, where the fuel cut flag M
FC is set to "0" and fuel cut is not executed.

Further, if it is judged to be the region III in the above step 131, it is judged to be a complete misfire, the process proceeds to step 144, the misfire judgment flag MFT is set to "1", and in the next step 145, the actual frequency [ a (%)] is set to 100%. Further, in the next step 146, the fuel cut flag MFC is set to "1" to execute the fuel cut.

In the embodiment (2) described above, the misfire determination method is switched in consideration of both the frequency 50% point [v50] of the ion current peak value [Pi] and the standard deviation value [σ]. Therefore, the ion current peak value [P
The misfire can be determined by an optimal misfire determination method that matches the distribution characteristic of i], and the misfire determination accuracy can be further improved.

In this embodiment (2), in the case of the region I, the ion current peak value [Pi] is the misfire determination value [Vth].
It is determined whether or not there is a misfire by determining whether the actual frequency [a (%)] calculated in step 107 of FIG. 5 is smaller than the second misfire determination value [C2]. The presence or absence of may be determined.

<< Embodiment (3) >> By the way, carbon is adhered to the ignition part insulator of the ignition plug 27 and the ignition plug 2
When the insulation resistance value between the electrodes 36 and 37 of No. 7 decreases, a phenomenon called "smolder" occurs in which a leakage current flows between the electrodes 36 and 37. When this smolder occurs, the smolder leakage current component is superimposed on the output of the ion current detection circuit 35 as shown in FIG.
Output is drifted due to smoldering leakage current component [c0
] It only shifts by a minute. If this drift value [c0] becomes large, the ion current peak value [Pi] may become larger than the misfire determination value [Vth] even when a misfire occurs, and the misfire detection accuracy will decrease.

Therefore, in the present embodiment (3), the output of the ion current detection circuit 35 is read as a drift value [c0] (zero point error) during the intake stroke (or compression stroke), and the ion current peak value [Pi ] The drift value [c0] is subtracted from the corrected ion current peak value [Pic] to be used as the combustion state detection value. These functions correspond to the drift value detection means and the output correction means in the claims.

FIG. 10 shows the relationship between the distribution characteristics of the ion current peak value [Pi] and the corrected ion current peak value [Pic] and the drift value [c0]. Figure 10
As shown in, when the smoldered state is stable, the fluctuation range of the drift value [c0] is small and the drift from the ion current peak value [Pi] occurs in both normal combustion and misfire occurrence. If the corrected ion current peak value [Pic] is obtained by subtracting the value [c0], this corrected ion current peak value [Pic] becomes a net ion current peak value without drift due to smoldering (zero point error). A misfire can be accurately detected from the ion current peak value [Pic].

On the other hand, when the smoldering state is unstable, the fluctuation range of the drift value [c0] becomes large and the standard deviation value [σc] or 50 of the drift value [c0] becomes 50, as shown in FIG. The percentage point [c50] also increases. In such a state, the drift value [c0] is subtracted from the ion current peak value [Pi] to obtain the corrected ion current peak value [Pi].
ic], the correction accuracy becomes poor and the misfire detection accuracy decreases.

Therefore, in this embodiment (3), the frequency 50% point [c50] and the standard deviation value [σc] in the distribution characteristic of the drift value [c0] are calculated, and the frequency 50% point [c]
50] and the standard deviation value [σc] are not within the predetermined range shown in FIG. 11, it is judged that the smoldering state is unstable and the misfire judgment is prohibited. The frequency 50% point [c50] and the standard deviation value [c50] The misfire determination is permitted only when σ c] is within the predetermined range shown in FIG.

The misfire determination of the present embodiment (3) described above is executed by the misfire determination routine of FIG. This misfire determination routine is executed for each combustion cycle. First, in step 201, the ion current peak value [Pi] and drift value [c0] are read, and in the next step 202,
It is determined whether or not the calculation of the distribution of the ion current peak value [Pi] and the drift value [c0] is permitted. If the calculation of the distribution is prohibited such as when the fuel is cut, the routine proceeds to step 204, where the data of the ion current peak value [Pi] and the drift value [c0] up to the previous time is stored.
The table and the c0 table are initialized and this routine ends.

On the other hand, if the calculation of the distribution is permitted, the routine proceeds to step 203, where the data of the ion current peak value [Pi] of this time is stored in the Pi table and the drift value [c0] of this time is stored in the c0 table. Store the data.

After that, the routine proceeds to step 205, where it is judged whether or not the number of data accumulated in the Pi table and the c0 table has exceeded a predetermined number (for example, 1000).
If the number of data is not equal to or larger than the predetermined number, this routine is terminated without performing the subsequent processing. And Pi
When the number of data stored in each of the table and the c0 table exceeds the predetermined number, the process proceeds to step 206,
The corrected ion current peak value [Pic] is obtained by subtracting the drift value [c0] from the ion current peak value [Pi].

Then, in the next step 207, the frequency 50% point [v50] and the standard deviation value [σ] in the distribution characteristic of the ion current peak value [Pi] are calculated, and the distribution characteristic of the drift value [c0] is calculated. Frequency at 50% point [c5
0] and the standard deviation value [σc] are calculated. After that, the routine proceeds to step 208, where the frequency 50% point [c50] and the standard deviation value [σc] in the distribution characteristic of the drift value [c0] are shown in FIG.
If the frequency 50% point [c50] and the standard deviation value [σc] are not within the predetermined range, the smoldering state is unstable and the ion current peak is When it is determined that the correction accuracy of the value [Pi] becomes poor, this routine is terminated without performing the subsequent processing. In this case, misfire determination is prohibited. This function plays a role corresponding to the misfire determination prohibition means in the claims.

On the other hand, when the frequency 50% point [c50] and the standard deviation value [σc] in the distribution characteristic of the drift value [c0] are within the predetermined range, it is judged that the smoldering state is stable, Proceeding to step 209, the frequency 50% point [v50] in the distribution characteristic of the corrected ion current peak value [Pic]
And the standard deviation value [σ], the distribution characteristic of the corrected ion current peak value [Pic] during normal combustion is predicted, and the predicted frequency at which the corrected ion current peak value [Pic] becomes equal to or less than the misfire determination value [Vth] [ A (%)] is calculated, and the actual frequency [a (%)] at which the actual corrected ion current peak value [Pic] is less than or equal to the misfire determination value [Vth] is calculated.
(%)] And the prediction frequency [A (%)] difference [Δ (%) =
a (%)-A (%)] is calculated.

After that, the routine proceeds to step 108 in FIG. 6, and the presence or absence of misfire is determined by the same processing as in the above-mentioned embodiment (1).

In the present embodiment (3) described above, the output of the ion current detection circuit 35 is read as the drift value [c0] (zero point error) during the intake stroke (or the compression stroke), and the ion current peak value [ Pi] to drift value [c0
] The corrected ion current peak value [Pi
Since c] is used as the combustion state detection value to determine the misfire, even if the output of the ion current detection circuit 35 drifts due to the smoldering leakage current, the output of the ion current detection circuit 35 will be the drift value [c0]. Correct with
The net ion current peak value without drift (zero point error) can be obtained, and the misfire determination accuracy can be improved.

Moreover, in the present embodiment (3), when the frequency 50% point [c50] and the standard deviation value [σc] in the distribution characteristic of the drift value [c0] are not within the respective predetermined ranges, the drift value [ Since the distribution characteristic of c0] is unstable and the correction accuracy of the ion current peak value [Pi] is deteriorated, the misfire determination is prohibited, so that the misdetection of misfire can be prevented.

In the present embodiment (1), misfire occurs depending on whether both the frequency 50% point [c50] and the standard deviation value [σc] in the distribution characteristic of the drift value [c0] are within a predetermined range. Although it is determined whether or not the determination is performed, it may be determined whether or not the misfire determination is performed based on only one of the frequency 50% point [c50] and the standard deviation value [σc]. << Embodiment (4) >> The above embodiments (1) to (3) are
Although the ion current peak value detected via the spark plug 27 is used as the combustion state detection value, in the embodiment (4) of the present invention shown in FIGS. 13 to 19, the cylinder pressure for detecting the cylinder pressure of the engine is used. A sensor (not shown) is attached to the engine, the cylinder pressure is detected by the cylinder pressure sensor at a plurality of timings, and the indicated mean effective pressure [Pe] is calculated based on these cylinder pressure detection values. This indicated mean effective pressure [P
e] is used as the combustion state detection value.

The indicated mean effective pressure [Pe] is calculated by the following equation. Pe = ∫ (ΔP) dV / Vo ΔP: In-cylinder pressure increase amount due to combustion (pressure difference from in-cylinder pressure at complete misfire shown in FIG. 13) dV: Stroke volume change amount at crank angle during in-cylinder pressure sampling Vo: total stroke volume equivalent to engine displacement

In the present embodiment (4), for example, BTDC1
In the crank angle range from 20 ° C to 120 ° C ATDC, for example, the in-cylinder pressure is sampled by the in-cylinder pressure sensor every 10 ° C, and the in-cylinder pressure increase amount due to combustion [Δ
P] is obtained by the pressure difference between the cylinder pressure in the compression stroke and the cylinder pressure in the expansion stroke (the cylinder pressure in the expansion stroke at the time of complete misfire is equal to the cylinder pressure in the compression stroke). For example, AT
Cylinder pressure increase due to combustion at DC 60 ° C [Δ
P] is BTDC60 from the cylinder pressure of ATDC60 ° C.
It can be obtained by subtracting the in-cylinder pressure of ° C A.

Further, in the present embodiment (4), offset correction (zero point correction), gain correction, and hysteresis correction are performed on the cylinder pressure sensor output [VP] (cylinder pressure detection value). As shown in FIG. 14, these corrections are carried out at specific crank angles [θ1, θ2] (for example, θ1 =
The in-cylinder pressure sensor outputs [VP1, VP2] of BTDC 60 ° C., θ2 = BTDC 20 ° C. A) are used.

The offset correction is to correct the zero point offset error (offset value) of the in-cylinder pressure sensor output [VP], for example, SAE2000-01-093.
The offset value [c0] of the in-cylinder pressure sensor output (see FIG. 15) is calculated by using the method described in 2.
It is calculated by a functional formula with P2 and the polytropic index as parameters. c0 = f (VP1, VP2, polytropic index)

Then, the offset value [c0] is subtracted from the in-cylinder pressure sensor output [VP] to perform offset correction of the in-cylinder pressure sensor output [VP]. VP = VP-c0 And the cylinder pressure sensor output after the offset correction [V
P] is used to perform gain correction and hysteresis correction.

On the other hand, the gain correction value [HG] is obtained from the ratio of the two using the in-cylinder pressure sensor outputs [VP1, VP2] at the specific crank angles [θ1, θ2] in the compression stroke. HG = VP2 / VP1

Then, the gain of the in-cylinder pressure sensor output [VP] is corrected by dividing the in-cylinder pressure sensor output [VP] by the gain correction value [HG]. VP = VP / HG

By this gain correction, it is possible to cancel the gain error for each in-cylinder pressure sensor and also to cancel the change in the in-cylinder pressure sensor output [VP] when the load changes. As a result, the cylinder pressure sensor output [V
The indicated mean effective pressure obtained as the combustion state detection value from P] also has a stable distribution characteristic that is not affected by the gain error or load change for each in-cylinder pressure sensor.

On the other hand, the hysteresis correction is to correct the hysteresis characteristic peculiar to the in-cylinder pressure sensor as shown in FIG. 17, and for example, the difference [HH] in the pressure detection sensitivities when the pressure increases and decreases. Is corrected by the difference between the rise and fall of the sampling value of the in-cylinder pressure sensor output [VP]. Specifically, the previous sampling value is VP (i-1),
Assuming that the sampled value this time is VP (i), when the pressure rises, VP (i) is hysteresis-corrected by the following equation. VP (i) = VP (i-1)-{VP (i-1) -VP (i)} / H
H

By this hysteresis correction, it is possible to correct the difference [HH] in the pressure detection sensitivities that differ when the pressure rises and when the pressure falls, and it is possible to accurately measure the in-cylinder pressure both when the pressure rises and when the pressure falls. Can be detected.

The misfire determination of the present embodiment (4) described above is executed by the misfire determination routine of FIGS. 18 and 19. This misfire determination routine is executed for each combustion cycle. First, in step 301, the cylinder pressure sensor output [V
P] is read, and in the next step 302, the indicated average effective pressure [Pe] is obtained using this cylinder pressure sensor output [VP].
Is calculated by the following formula. Pe = ∫ (ΔVP) dV / Vo

Here, ΔVP is the amount of increase in the in-cylinder pressure sensor output due to combustion (in-cylinder pressure increase amount), and is determined by the difference between the in-cylinder pressure sensor output in the compression stroke and the in-cylinder pressure sensor output in the expansion stroke. Ask. For example, in-cylinder pressure sensor output increase amount [ΔVP] due to combustion at ATDC 60 ° C.
Is the BTD from the output of the in-cylinder pressure sensor at ATDC 60 ° C.
It is obtained by subtracting the output of the in-cylinder pressure sensor at C60 ° C.

After this, the routine proceeds to step 303, where, for example, S
Using the method described in AE2000-01-0932, the offset value [c0] of the in-cylinder pressure sensor output
(See FIG. 15) is calculated by a functional expression having VP1, VP2 and the polytropic index as parameters. c0 = f (VP1, VP2, polytropic index)

Then, in the next step 304, using this offset value [c0], the cylinder pressure sensor output [V
P] offset correction is performed. VP = VP-c0 After that, the routine proceeds to step 305, where the in-cylinder pressure sensor output [VP1 at a specific crank angle [θ1, θ2] in the compression stroke is obtained.
, VP2], the gain correction value [HG] is calculated by the following equation. HG = VP2 / VP1

Then, in the next step 306, the gain correction value [HG] is used to correct the gain of the in-cylinder pressure sensor output [VP] by the following equation. VP = VP / HG

After that, the process proceeds to step 307, and the difference [HH] in the pressure detection sensitivities of the in-cylinder pressure sensor is obtained from the difference between the rising and falling of the sampling value of the in-cylinder pressure sensor output [VP]. Then, in the next step 308, hysteresis correction of the in-cylinder pressure sensor output [VP] is performed by a predetermined functional expression using the pressure detection sensitivity difference [HH] of the in-cylinder pressure sensor. The processing of these steps 303 to 308 serves as a correcting means in the claims.

After that, the routine proceeds to step 309, where the indicated average effective pressure [P] is obtained using the in-cylinder pressure sensor output [VP] which has been offset-corrected, gain-corrected and hysteresis-corrected.
eV] is used as the combustion state detection value and calculated by the following equation. PeV = ∫ (ΔVP) dV / Vo

Then, the process proceeds to step 310 in FIG.
It is determined whether calculation of the distribution of the combustion state detection value [PeV] and the offset value [c0] is permitted. If the distribution calculation is prohibited, such as at the time of fuel cut, the routine proceeds to step 312, where the combustion state detection value [PeV] up to the previous time is reached.
Then, the PeV table and the c0 table which store the data of the offset value [c0] and the table are initialized, and this routine is finished.

On the other hand, if the calculation of the distribution is permitted, the process proceeds to step 311, and the data of the combustion state detection value [PeV] of this time is stored in the PeV table, and the offset value [c0] of this time is stored in the c0 table. Store the data.

Then, the process proceeds to step 313, and it is judged whether or not the number of data stored in the PeV table and the data stored in the c0 table have exceeded a predetermined number (for example, 1000).
If the number of data is not equal to or larger than the predetermined number, this routine is terminated without performing the subsequent processing. And PeV
When the number of data stored in each of the table and the c0 table exceeds the predetermined number, the process proceeds to step 314.
Frequency 50% in the distribution characteristics of the combustion state detection value [PeV]
The point [v50] and the standard deviation value [σ] are calculated, and the frequency 50% point [c50] and the standard deviation value [σc] in the distribution characteristic of the offset value [c0] are calculated.

After that, the process proceeds to step 315, and the frequency 50% point [c50] in the distribution characteristic of the offset value [c0].
And standard deviation value [σc] are within a predetermined range, and the frequency 50% point [c50] and standard deviation value [σc]
Is not within the predetermined range, the combustion state detection value [Pe
V] is determined to be inaccurate, and the combustion state detection value [Pe
Misfire judgment based on the distribution characteristics of V] is prohibited.

In this case, the routine proceeds to step 321, where the indicated mean effective pressure before correction [P
e] is compared with a predetermined misfire determination value [Vth], and if the indicated mean effective pressure [Pe] is smaller than the misfire determination value [Vth],
When it is determined that a misfire has occurred, the routine proceeds to step 322, where the misfire determination flag MFT is set to "1". On the other hand, if the indicated mean effective pressure [Pe] is equal to or higher than the misfire determination value [Vth], it is determined that there is no misfire, the process proceeds to step 323, and the misfire determination flag MFT is set to "0".

On the other hand, if it is determined in step 315 that the frequency 50% point [c50] and the standard deviation value [σc] in the distribution characteristic of the offset value [c0] are within the predetermined range, combustion state detection is performed. When it is judged that the correction accuracy of the value [PeV] is good, the routine proceeds to step 316, where the combustion state detection value [PeV]
Based on the frequency 50% point [v50] and standard deviation [σ] in the distribution characteristic of V], the combustion state detection value during normal combustion [PeV]
Of the actual combustion state detection value [PeV] while calculating the prediction frequency [A (%)] at which the combustion state detection value [PeV] becomes equal to or less than the misfire determination value [Vth] by predicting the distribution characteristic of Calculate the actual frequency [a (%)] that is less than or equal to the value [Vth],
A difference [Δ (%) = a (%) − A (%)] between the actual frequency [a (%)] and the predicted frequency [A (%)] is calculated.

After that, the routine proceeds to step 108 in FIG. 6, and the presence or absence of misfire is judged by the same processing as that of the above-mentioned embodiment (1). Also in the present embodiment (4) described above, it is possible to accurately determine the presence or absence of misfire based on the distribution characteristics of the combustion state detection value [PeV].

In the present embodiment (4), the indicated mean effective pressure is calculated based on the in-cylinder pressure sensor output, and this indicated mean effective pressure is used as the combustion state detection value. Instead of the pressure, any one of the work rate, the heat generation rate, and the work amount may be calculated as the combustion state detection value.

Here, the indicated mean effective pressure is Pe (N /
m ² ), the plane area of the piston top surface is S (m ² ), and the stroke of the piston is L (m), the work from the top dead center to the bottom dead center of the expansion stroke per cylinder is It is calculated by the formula. Work = Pe x S x L

Further, in the case of a 4-cycle engine, there is an expansion stroke at a rate of once every two revolutions. Therefore, assuming that the engine rotation speed is N (rpm), the power is calculated by the following equation. Work rate = 1/2 × N / 60 × Work amount = 1/2 × N / 60 × Pe × S × L

Further, in calculating the heat generation rate, a specific crank angle [θ3, θ4] of the expansion stroke (eg, θ3 = ATD) is calculated.
Using the in-cylinder pressure sensor outputs [VP3, VP4] at C20 ° C, θ4 = ATDC60 ° CA), the heat generation rate is calculated by the following equation. Heat generation rate = VP3 / VP4

<< Embodiment (5) >> In the present embodiment (5), the indicated mean effective pressure [PeV] is calculated as the combustion state detection value by the same method as in the above embodiment (4), and the compression stroke is further calculated. The in-cylinder pressure sensor output read at the predetermined crank angle is set as the reference output [VPbase], and the combustion state detection value [P
eV] divided by the reference output VPbase [PeV / VPba
The presence or absence of misfire is determined by whether or not se] is smaller than the misfire determination value [Vth].

For example, in step 315 of FIG. 19, the frequency 50% point [c] in the distribution characteristic of the offset value [c 0].
50] and the standard deviation value [σc] are not within the predetermined range, the value [PeV] obtained by dividing the combustion state detection value [PeV] by the reference output [VPbase] instead of the process of step 321. Whether or not there is a misfire is determined by whether or not / VPbase] is smaller than the misfire determination value [Vth]. And PeV / V
If Pbase is smaller than the misfire determination value [Vth], it is determined that a misfire has occurred, the misfire determination flag MFT is set to "1", and if the indicated mean effective pressure [Pe] is greater than or equal to the misfire determination value [Vth]. , Misfire determination flag MFT
Is set to "0" (steps 322 and 323).

The combustion state detection value [PeV] and the reference output [VPbase] may or may not be offset-corrected, gain-corrected, and hysteresis-corrected.

<< Embodiment (6) >> The foregoing embodiments (1) to
A configuration may also be used in which the means for determining misfire from the peak value of the ion current by any of the methods of (3) and the means for determining misfire from the output of the in-cylinder pressure sensor by the method of the embodiment (4) or (5) are used together. good. In this case, depending on the engine operating state and the like, the misfire determination means having a higher misfire determination accuracy may be selected to perform the misfire determination, or the two misfire determination means may simultaneously perform the misfire determination to determine the two misfire determinations. Only when both of the means determine that a misfire has occurred, it may be possible to finally determine that a misfire has occurred.

<< Embodiment (7) >> By the way, in each of the above embodiments (1) to (6), the misfire determination value [Vth] is a preset fixed value for simplification of the calculation process. As shown in FIG. 20, the ion current peak value [Pi] detected by the ion current detection circuit 35 may vary depending on the presence or absence of smoldering of the spark plug 27, variations in the combustion state, and the like, so the misfire determination value [Vth]. Initial setting value [V1]
When fixed by, the ion current peak value [Pi
], There is a concern that the misfire cycle may be erroneously determined as a normal combustion cycle [see FIG. 20 (b)].

Therefore, in the embodiment (7) of the present invention, the misfire determination routine of FIG. 23 is executed for each combustion cycle, and the misfire determination value setting routine of FIG. 24 is executed at step 400 to detect the ion current. The misfire determination value [Vth] is set based on the distribution characteristic of the ion current peak value [Pi] detected by the circuit 35.

Here, the misfire determination method of this embodiment (7) will be described with reference to FIGS. Misfire judgment value [Vt
When setting the initial setting value [V1] of h],
As shown in (a), the distribution pattern of the misfire cycle was measured in advance by an experiment or a simulation without the smoldering of the spark plug 27, and the ion current peak value [Pi] belonging to this distribution pattern of the misfire cycle was measured. The maximum ion current peak value of 1 or a value slightly larger than that is set as the initial setting value [V1].

Further, a preliminary misfire determination value [V2] is set in the vicinity of an intermediate position between the misfire cycle distribution pattern and the normal combustion cycle distribution pattern by experiments or simulations in advance. This provisional misfire determination value [V2] is
As shown in FIG. 20 (b) and FIG. 21, even when smoldering of the spark plug 27 occurs, the misfire cycle distribution pattern is set so as not to overlap the temporary misfire determination value [V2]. Therefore, as shown in FIG. 20 (a), when there is no smolder, the provisional misfire determination value [V2] exists at a position far away from the misfire cycle distribution pattern.

Next, each parameter for evaluating the distribution characteristic of the ion current peak value [Pi] will be described with reference to FIG. NS is the height of the distribution pattern of the misfire cycle (maximum detection frequency), and the ion current detected at the highest frequency [N] within the range of 0 ≦ Pi ≦ V2, which is the range including the distribution pattern of the misfire cycle. It is the detection frequency of the peak value [PS].

NB is the height of the distribution pattern of the normal combustion cycle (maximum detection frequency), and is the highest frequency [N] within the range of V2 <Pi, which is the range including the distribution pattern of the normal combustion cycle. It is the detection frequency of the detected ion current peak value [PB]. In addition, these PB and PS are
In order to statistically process the distribution characteristic of the ion current peak value [Pi], it is represented by being converted into a division section value having a predetermined width. In case of misfire in all cycles (NB = 0), PB
Is set to the maximum division value.

Further, H is a region where the detection frequency [N] is smaller than a predetermined value (for example, 1) between the maximum detection frequency positions [PS] and [PB] of both distribution patterns (that is, misfire cycle and normal combustion). The width of the noise region between both distribution patterns of the cycle), which is expressed in terms of the number of divided sections.

Po is a value slightly larger than the maximum ion current peak value of the ion current peak values [Pi] belonging to the misfire cycle distribution pattern, and corresponds to a new misfire determination value described later. Then, it is expressed by being converted into a division classification value. When calculating Po, it is within the range between the division classification value [PS] of the maximum detection frequency position of the distribution pattern of the misfire cycle and the temporary misfire determination value [V2] (PS <Pi <V2
), The minimum ion current peak value in a region (noise region) where the detection frequency [N] is smaller than a predetermined value (for example, 1) is obtained as a new misfire determination value [Po].

In this case, the degree of separation [f] between the misfire cycle distribution pattern and the normal combustion cycle distribution pattern is
It is calculated by the following equation using the width [H] of the noise region between both distribution patterns and the height parameter [W] of both distribution patterns. f = H × W

Here, the height parameter [W] of both distribution patterns is the maximum detection frequency [NS] of both distribution patterns,
It is a value obtained by adding [NB]. W = NS + NB The degree of separation [f] of both distribution patterns means that the larger the value is, the more clearly the distribution patterns are separated.

As shown in FIG. 22, during unstable combustion,
Both distribution patterns of the normal combustion cycle and the misfire cycle overlap, and the width of the noise region between both distribution patterns [H]
Becomes 0, the separation [f] also becomes 0.

In the present embodiment (7), it is determined whether or not both distribution patterns are clearly separated based on the degree of separation [f] of both distribution patterns and the height parameter [W]. When it is determined that the components are clearly separated, a new misfire determination value [Po] is calculated based on the distribution pattern of the misfire cycle. If it is determined that the two distribution patterns are not clearly separated, the misfire determination value [Po] is not calculated, and the stored value of the previous misfire determination value [Po] is maintained as it is.

Then, as described in the above embodiment (3), considering that the ion current peak value [Pi] detected by the ion current detection circuit 35 drifts due to the smoldering of the ignition plug 27, the above embodiment is adopted. If the drift value [c0] of the ion current peak value [Pi] detected by the same method as (3) is not less than the predetermined value [K3], the misfire determination value [Vth] is determined based on the distribution pattern of the misfire cycle. The calculated new misfire determination value [Po] is used. On the contrary, if the drift value [c0] is smaller than the predetermined value [K3], the misfire determination value [Vth] is set to the initial setting value [V1].
To use.

The setting of the misfire determination value [Vth] described above is executed by the misfire determination value setting routine of FIG. 24 which is started in step 400 of the misfire determination routine of FIG. Incidentally, step 400 of the misfire determination routine of FIG.
Since the processing of each step other than is the same as the processing of each step of the misfire determination routine of FIG. 12 described in the above embodiment (3), description thereof will be omitted.

Step 40 of the misfire determination routine of FIG.
The misfire determination value setting routine of FIG. 24, which is started at 0, serves as a misfire determination value calculating means in the claims. When this routine is started, first, step 40
In 1, the height parameter [W] is obtained by adding the maximum detection frequencies [NS] and [NB] of both normal combustion cycle and misfire cycle distribution patterns. W = NS + NB

Thereafter, the routine proceeds to step 402, where the separation degree [f] of both distribution patterns of the normal combustion cycle and the misfire cycle is set to the width [H] of the noise region between both distribution patterns and the height parameter of both distribution patterns [f]. W] and the following formula. f = H × W

At this time, the width [H] of the noise region between both distribution patterns is determined by the maximum detection frequency position [P] of both distribution patterns.
The width of the region where the detection frequency [N] is smaller than a predetermined value (for example, 1) between S] and [PB] may be calculated and obtained.

Then, in the next step 403, it is determined whether or not the condition for calculating the misfire determination value is satisfied by the following two conditions,
It is determined by whether or not both are satisfied. Separability [f] of both distribution patterns is equal to or greater than a predetermined value [K1] (f ≧ K1) Height parameter [W] of both distribution patterns is a predetermined value [K]
2] or more (H ≧ K2)

If any one of these two conditions is not satisfied, the condition for calculating the misfire determination value is not satisfied, and the calculation of the misfire determination value [Po] in step 404 is not performed. move on.

On the other hand, if both of the above two conditions are satisfied, the misfire determination value calculation condition is satisfied, and the routine proceeds to step 404, where a new misfire determination value [Po] is obtained based on the distribution pattern of the misfire cycle. The calculated value of the misfire determination value [Po] stored in the memory (storage means) of the microcomputer 41 is updated. This new misfire determination value [Po] corresponds to a value slightly larger than the maximum corrected ion current peak value of the corrected ion current peak values [Pic] belonging to the distribution pattern of the misfire cycle, and is calculated. In this case, the division classification value [PS] of the peak position of the distribution pattern of the misfire cycle and the temporary misfire determination value [V2]
Within the range (PS <Pic <V2), the minimum corrected ion current peak value in a region (noise region) where the detection frequency [N] is smaller than a predetermined value (for example, 1) is set as a new misfire determination value [Po]. You can ask for The corrected ion current peak value [Pic] is obtained by subtracting the drift value [c0] from the ion current peak value [Pi] in step 206. The process of step 403 serves as a permission / prohibition determination means in the claims.

As described above, the new misfire determination value [P
o]], the process proceeds to step 405, where the drift value [c0] of the ion current peak value [Pi] is a predetermined value [K3].
] Or more, and if the drift value [c0] is greater than or equal to the predetermined value [K3], the process proceeds to step 406,
The new misfire determination value [P
o] is selected as the misfire determination value [Vth] to be used this time. When the step 404 is skipped (when the misfire determination value [Po] is not calculated), the stored value of the misfire determination value [Po] calculated in step 404 when this routine was started in the past. Should be used.

On the other hand, if the drift value [c0] of the ion current peak value [Pi] is smaller than the predetermined value [K3], the process proceeds to step 407, where the initial set value [V1] is used as the misfire determination value [Vth]. To choose as. The processing of steps 405 to 407 serves as a misfire determination value selecting means in the claims.

After setting the misfire determination value [Vth] as described above, the process proceeds to step 209 in FIG. 23, where the frequency of the corrected ion current peak value [Pic] in the distribution characteristic is 50%.
The distribution characteristic of the corrected ion current peak value [Pic] during normal combustion is predicted from the point [v50] and the standard deviation value [σ], and the corrected ion current peak value [Pic] becomes the misfire determination value [Vth].
The predicted frequency [A (%)] which will be described below is calculated, and the actual corrected ion current peak value [Pic] is set to the misfire determination value [V
The actual frequency [a (%)] that is less than or equal to th] is calculated, and the difference [Δ between the actual frequency [a (%)] and the predicted frequency [A (%)]
(%) = A (%) − A (%)] is calculated.

After that, the routine proceeds to step 108 in FIG. 6, and the presence or absence of misfire is determined by the same processing as in the above-mentioned embodiment (1).

According to the embodiment (7) described above,
Since the misfire determination value [Vth] is set based on the distribution characteristic of the ion current peak value [Pi], the distribution of the ion current peak value [Pi] depends on the presence or absence of smoldering of the spark plug 27, the variation of the combustion state, and the like. Even if the characteristics deviate, the misfire determination value [Vth] can be set in consideration of the deviation, and the deviation of the ion current peak value [Pi] due to the presence or absence of smoldering of the ignition plug 27, the variation of the combustion state, and the like. The misfire cycle and the normal combustion cycle can be accurately discriminated with little influence, and a highly reliable misfire determination can be performed for a long period of time.

Moreover, in the present embodiment (7), the separability [f] and the height parameter [W] of both distribution patterns, which are parameters showing the distribution characteristics of the ion current peak value [Pi].
Since it is determined whether to allow or prohibit the calculation of the new misfire determination value [Po] based on the above, only when the variation of the ion current peak value [Pi] is small (that is, between the normal combustion cycle and the misfire cycle). Only when both distribution patterns are clearly separated), the misfire determination value [Vth] is calculated. When the ion current peak value [Pi] has a large variation, the misfire determination value [Vth] need not be calculated. It is possible to prevent the accuracy of misfire determination from being deteriorated due to the variation of the peak value [Pi] of the ion current.

Further, in the present embodiment (7), the drift value [c0] of the ion current peak value [Pi] is the predetermined value [K3.
] The value of the misfire judgment value used this time [Vt
As h], it is determined whether a new misfire determination value [Po] calculated based on the distribution characteristic of the ion current peak value [Pi] or an initial setting value [V1] is selected. The optimum misfire determination value [Vth] can be set according to the presence or absence of smolder, and highly reliable misfire determination that is not affected by the presence or absence of smolder can be performed.

When calculating a new misfire determination value [Po] based on the distribution characteristic of the ion current peak value [Pi], a new misfire determination is made for each engine operating condition or for each degree of smoldering (for each drift value). The value [Po] is calculated and stored in the memory, and depending on the engine operating conditions or the degree of smoldering,
An appropriate misfire determination value [Po] may be selected from the stored values in the memory.

The distribution pattern of the ion current peak value [Pi] measured in the fuel cut region during deceleration is regarded as the distribution pattern of the misfire cycle, and a new misfire determination value [Po] is determined based on the distribution pattern of the misfire cycle. May be calculated and stored in the memory, and the misfire determination value [Po] stored in the memory after the fuel cut is restored may be read and used.

In the present embodiment (7), a new misfire determination value is obtained based on the distribution characteristic of the corrected ion current peak value [Pic] obtained by subtracting the drift value [c0] from the ion current peak value [Pi]. Although the misfire determination is performed by calculating [Po], a new misfire determination value [Po] is calculated based on the distribution characteristic of the ion current peak value [Pi] without subtracting the drift value [c0]. You may make it misfire determination.

Further, in the present embodiment (7), the ion current peak value [Pi] detected by using the spark plug 27 is used as the combustion state detection value, but other misfire detection such as the cylinder pressure sensor is detected. Using the sensor, a new misfire determination value [Po
], And misfire determination may be performed.

Further, in the present embodiment (7), it is determined whether the calculation of the new misfire determination value [Po] is permitted or prohibited based on the distribution characteristic of the ion current peak value [Pi]. Alternatively, it may be determined whether to permit or prohibit the calculation of the new misfire determination value [Po] based on the engine operating state. For example, when the engine operating state in which the ion current output is stable, it is determined that the distribution patterns of both the normal combustion cycle and the misfire cycle are clearly separated, and calculation of a new misfire determination value [Po] is permitted. The calculation of the new misfire determination value [Po] may be prohibited when the engine operating state in which the ion current output is not stable.

Further, in the present embodiment (7), the maximum detection frequencies [NS] and [NB] of both distribution patterns of the normal combustion cycle and the misfire cycle are obtained, but the center of gravity of each predetermined output distribution band is determined. The distribution pattern determination method may be changed as appropriate, such as the position may be obtained.

In the present embodiment (7), the misfire determination value [Po] calculated based on the distribution characteristic of the ion current peak value [Pi] and the initial set value [V1] are used as the ion current peak value [Pi]. Although the selection was made according to the drift value [c0] of the engine, the misfire judgment value [P
o] and the initial setting value [V1] may be selected.

Alternatively, the function of calculating a new misfire determination value [Po] based on the distribution characteristic of the ion current peak value [Pi] is omitted, and experiments or simulations are performed in advance for each engine operating state and / or ion. Current peak value [Pi]
A plurality of misfire determination values [Vth] are prepared and stored in a memory for each distribution characteristic, and the plurality of misfire determination values [Vth] are stored according to the engine operating condition and / or the ion current peak value [Pi] distribution characteristics. The misfire determination value [Vth] used this time may be selected from Vth]. Even in this case, the optimum misfire determination value [Vth] can be selected from the plurality of misfire determination values [Vth] according to the presence or absence of smoldering of the ignition plug 27, the variation in the combustion state, and the like. It is possible to perform highly reliable misfire determination that is less affected by the presence or absence of smoldering of the plug 27 and the variation of the combustion state.

[Brief description of drawings]

FIG. 1 is a circuit diagram showing a configuration of an ignition control system and an ion current detection circuit in an embodiment (1) of the present invention.

FIG. 2 is a time chart showing a signal waveform of each part of the embodiment (1).

FIG. 3 is a diagram for explaining distribution characteristics of the ion current peak value [Pi] during normal combustion according to the embodiment (1).

FIG. 4 is a diagram for explaining changes in distribution characteristics of the ion current peak value [Pi] during normal combustion and when misfire occurs in the embodiment (1).

FIG. 5 is a flowchart showing the flow of processing in the first half of the misfire determination routine of the embodiment (1).

FIG. 6 is a flowchart showing a processing flow of the latter half of the misfire determination routine of the embodiment (1).

FIG. 7 is a diagram conceptually showing a misfire determination method switching map used in the embodiment (2).

FIG. 8 is a flowchart showing a processing flow of the latter half of the misfire determination routine of the embodiment (2).

FIG. 9 is a time chart showing a signal waveform of each part of the embodiment (3).

FIG. 10 is a diagram for explaining changes in the distribution characteristic of the ion current peak value [Pi] during normal combustion and when misfire occurs in the embodiment (3).

FIG. 11 is a view conceptually showing a misfire determination permission / prohibition switching map of the embodiment (3).

FIG. 12 is a flowchart showing the flow of processing in the first half of the misfire determination routine of the embodiment (3).

FIG. 13 is a time chart showing a change characteristic of in-cylinder pressure detected in the embodiment (4).

FIG. 14 is a time chart showing a change characteristic of an in-cylinder pressure sensor output.

FIG. 15 is a diagram illustrating offset correction of in-cylinder pressure sensor output.

FIG. 16 is a diagram for explaining gain correction of an in-cylinder pressure sensor output.

FIG. 17 is a diagram for explaining hysteresis correction of in-cylinder pressure sensor output.

FIG. 18 is a flowchart showing the flow of processing in the first half of the misfire determination routine of the embodiment (4).

FIG. 19 is a flowchart showing a processing flow of the latter half of the misfire determination routine of the embodiment (4).

20A is a diagram illustrating the distribution characteristic of the ion current peak value [Pi] without smoldering, and FIG. 20B is a diagram illustrating the distribution characteristic of the ion current peak value [Pi] when smoldering occurs.

FIG. 21: Ion current peak value [Pi when smoldering occurs]
For explaining the method of setting the misfire determination value [Vth] using the distribution characteristic of

FIG. 22 is a diagram for explaining distribution characteristics of ion current peak values [Pi] during unstable combustion without smoldering.

FIG. 23 is a flowchart showing the flow of processing of the first half of the misfire determination routine of the embodiment (7).

FIG. 24 is a flowchart showing a processing flow of a misfire determination value setting routine of the embodiment (7).

[Explanation of symbols]

21 ... Ignition coil, 22 ... Primary coil, 23 ... Battery, 24 ... Igniter, 25 ... Power transistor, 2
6 ... Secondary coil, 27 ... Spark plug, 31 ... Ion current detection resistor, 33 ... Inversion amplification circuit, 34 ... Engine control circuit, 35 ... Ion current detection circuit (combustion state detection means),
36 ... Central electrode, 37 ... Ground electrode, 38 ... Noise mask, 39 ... Peak hold circuit, 41 ... Microcomputer (distribution characteristic evaluation means, misfire determination means, drift value detection means, output correction means, misfire determination prohibition means, misfire) Judgment value calculation means, permission / prohibition judgment means, misfire judgment value selection means).

─────────────────────────────────────────────────── ─── Continuation of front page (51) Int.Cl. ⁷ Identification code FI theme code (reference) F02P 17/00 F (72) Inventor Koichi Nakata 1 Toyota-cho, Toyota-shi, Aichi Prefecture Toyota Motor Corporation ( 72) Inventor Kazuhisa Mogi 1 Toyota Town, Toyota City, Aichi Prefecture, Toyota Motor Co., Ltd. (72) Inventor, Akito Onishi 1 Toyota Town, Toyota City, Aichi Prefecture, Toyota Motor Co., Ltd. F Term (reference) 3G019 CD01 CD06 DB07 DB08 DB14 3G084 DA27 EA11 EC04 FA00 FA21 FA24

Claims (21)

[Claims] 1. A combustion state detecting means for detecting a combustion state of an internal combustion engine, and a combustion state detection value for each combustion cycle detected by the combustion state detecting means is statistically processed to evaluate a distribution characteristic of the combustion state detection value. And a distribution characteristic evaluation means for determining the presence or absence of misfire by determining whether the distribution characteristic of the combustion state detection value evaluated by the distribution characteristic evaluation means is the distribution characteristic at the time of normal combustion or the distribution characteristic at the time of misfire occurrence. A misfire detecting device for an internal combustion engine, comprising: 2. The combustion state detecting means detects an ion current generated by combustion in a cylinder, and uses the ion current peak value as the combustion state detection value. The distribution characteristic determining means determines the ion current. The misfire detection device for an internal combustion engine according to claim 1, wherein a statistically processed value of a logarithmic normal distribution of the peak value is obtained as a distribution characteristic value of the combustion state detection value. 3. The distribution characteristic evaluation means calculates a frequency 50% point [v50] and a standard deviation value [σ] in the distribution characteristic of the combustion state detection value, and the frequency 50% point [v50]. Means for predicting the distribution characteristic of the combustion state detection value at the time of normal combustion and the prediction frequency [A (%)] at which the combustion state detection value becomes equal to or less than the misfire determination value [Vth] And the actual combustion state detection value is the misfire determination value [Vt
h] or less, a means for obtaining an actual frequency [a (%)] and a difference [Δ (%) = a (%) between the actual frequency [a (%)] and the predicted frequency [A (%)] -A (%)], and the misfire determination means determines the presence / absence of misfire based on the difference [Δ (%)].
A misfire detection device for an internal combustion engine according to item 1. 4. The misfire determination means is configured to detect the difference [Δ
4. The misfire detection device for an internal combustion engine according to claim 3, further comprising means for calculating a misfire occurrence rate based on (%). 5. The misfire determination means is configured to detect the difference [Δ
(%)] For determining the presence or absence of misfire, and means for determining the presence or absence of misfire based on the actual frequency [a (%)] or the combustion state detection value. The misfire detection device for an internal combustion engine according to claim 3 or 4, wherein switching is performed according to a distribution characteristic value such as a percentage point [v50]. 6. The misfire determination means is in a state where a stable combustion state detection value distribution characteristic is obtained in which each of the frequency 50% point [v50] and the standard deviation value [σ] is within a predetermined range. The misfire detection device for an internal combustion engine according to claim 3 or 4, wherein the presence or absence of misfire is determined based on the difference [Δ (%)] at times. 7. A drift value detection means for detecting a drift value of the output of the combustion state detection means, and a combustion state detection value obtained by correcting the output of the combustion state detection means with the drift value detected by the drift value detection means. And a distribution characteristic evaluation means for statistically processing the combustion state detection value for each combustion cycle corrected by the output correction means to evaluate the distribution characteristic of the combustion state detection value. The misfire detection device for an internal combustion engine according to any one of claims 1 to 6. 8. A means for calculating a frequency 50% point [c50] and a standard deviation value [σc] in the drift value distribution characteristic detected by the drift value detecting means, and a frequency 50% in the drift value distribution characteristic. Point [c5
0] and the standard deviation value [σc] are not within predetermined ranges, respectively, and misfire determination prohibiting means for prohibiting the misfire determination by the misfire determining means.
A misfire detection device for an internal combustion engine according to item 1. 9. The misfire of an internal combustion engine according to claim 7, wherein the drift value detecting means detects an output of the combustion state detecting means during an intake stroke or a compression stroke as a drift value. Detection device. 10. The combustion state detecting means detects the in-cylinder pressure at a plurality of timings, and based on the in-cylinder pressure detection values, any one of the work rate, the heat generation rate, the indicated mean effective pressure, and the work amount. 2. The misfire detection device for an internal combustion engine according to claim 1, wherein is calculated as the combustion state detection value. 11. The combustion state detecting means includes a correcting means for correcting at least one of an offset correction, a gain correction, and a hysteresis correction with respect to the in-cylinder pressure detection value. Item 11. A misfire detection device for an internal combustion engine according to item 10. 12. The misfire determination value [Vt based on the distribution characteristic of the combustion state detection value evaluated by the distribution characteristic evaluation means.
and a misfire determination value calculating means for calculating h], wherein the misfire determination means determines the presence or absence of a misfire based on the distribution characteristic of the combustion state detection values and the misfire determination value [Vth]. A misfire detection device for an internal combustion engine according to any one of claims 1 to 6. 13. The misfire determination value calculation means separates a distribution characteristic pattern of combustion state detection values evaluated by the distribution characteristic evaluation means into a normal combustion cycle distribution pattern and a misfire cycle distribution pattern, The misfire detection device for an internal combustion engine according to claim 12, wherein the misfire determination value [Vth] is calculated based on a misfire cycle distribution pattern. 14. The misfire determination value calculation means sets the maximum combustion state detection value among the combustion state detection values belonging to the distribution pattern of the misfire cycle or a value slightly larger than the maximum combustion state detection value as the misfire determination value [Vth]. Claim 1 characterized by the above.
The misfire detection device for an internal combustion engine according to item 3. 15. The misfire determination value [Vth] is allowed to be calculated by the misfire determination value calculation means based on the operating state of the internal combustion engine and / or the distribution characteristic of the combustion state detection value evaluated by the distribution characteristic evaluation means. 3. A permission / prohibition determination means for determining whether to prohibit or prohibit.
The misfire detection device for an internal combustion engine according to any one of 2 to 14. 16. The misfire determination value [Vth] by the misfire determination value calculation means when the separation degree between the normal combustion cycle distribution pattern and the misfire cycle distribution pattern is a predetermined value or more. 16. The misfire detection device for an internal combustion engine according to claim 15, wherein the calculation is allowed. 17. The permission / prohibition determination means determines the degree of separation between the normal combustion cycle distribution pattern and the misfire cycle distribution pattern based on the width of the noise region between both distribution patterns and the height of both distribution patterns. The misfire detection device for an internal combustion engine according to claim 16, wherein the determination is made. 18. A plurality of misfire determination values [Vt including a latest misfire determination value [Vth] calculated by the misfire determination value calculating means.
[hth] is stored, and based on the operating condition of the internal combustion engine and / or the distribution characteristic of the combustion state detection value evaluated by the distribution characteristic evaluating means, the misfire determination value [Vth] is used this time. The misfire detection device for an internal combustion engine according to any one of claims 12 to 17, further comprising: a misfire determination value selection unit that selects a misfire determination value [Vth]. 19. The misfire determination value selection means selects an initial setting value from the plurality of misfire determination values [Vth] when the drift value of the output of the combustion state detection means is smaller than a predetermined value. The misfire detection device for an internal combustion engine according to claim 18, wherein: 20. The misfire determination value selection means selects the latest misfire determination value [Vth] calculated by the misfire determination value calculation means when the drift value of the output of the combustion state detection means is a predetermined value or more. The misfire detection device for an internal combustion engine according to claim 18 or 19, wherein: 21. A misfire determination value [Vth] to be used this time from among a plurality of misfire determination values [Vth] based on an operating state of an internal combustion engine and / or a distribution characteristic of combustion state detection values evaluated by the distribution characteristic evaluation means. The misfire detection device for an internal combustion engine according to any one of claims 18 to 20, further comprising: misfire determination value selection means for selecting.