DE10052140B4 - Combustion conditioner for engines - Google Patents

Abstract

Combustion condition controller for an internal combustion engine with:
a torque change calculating means (203) for calculating a torque variation of an internal combustion engine (10) for each combustion; and
a roughness calculating means (303 to 307) for calculating a combustion roughness serving as a parameter for displaying the combustion state by statistically processing the torque variation of an effective predetermined sample inventory parameter under a condition that only data in a predetermined established range of the calculated torque variation is considered to be effective (303) using only a predetermined number (n) of these data considered to be effective (305, 306).

Description

These The invention relates to a device for detecting and controlling or rules of a combustion state of an internal combustion engine and in particular, it relates to a device for detecting and controlling or rules of combustion roughness, as a parameter to indicate the combustion state.

Heretofore, a control apparatus is known for controlling an internal combustion engine that detects the combustion state from a rotation change for each combustion in the internal combustion engine and reflects the detected result on the air-fuel ratio. In detail, for example, in a six-cylinder internal combustion engine, the engine speed changes every 120 ° crank angle, as in FIG 10 is shown. Further, in detail, the rotational speed reaches the maximum value and the minimum value at each combustion. The rotational change ΔNe is calculated from the difference between the maximum value and the minimum value of the rotational speed in each combustion, and the combustion state is detected based on the rotational variation. In the true operation, the combustion roughness is calculated based on the standard deviation of the rotational change, and the combustion state is evaluated based on the calculated result.

It It is known that the dispersion of the combustion state is more elevated, if the air-fuel ratio is shifted to the lean side, and the combustion roughness increases, which serves as the parameter. When the air-fuel ratio at the lean range is regulated, the air-fuel ratio is so regulated that the combustion state is not the tolerance limit driveability in monitoring exceeds the combustion roughness.

The distribution result of the obtained rotation change measurement in the change of the air-fuel ratio from the stoichiometric air-fuel ratio to the lean region is in FIG 11a to 11c shown. As in 11a is shown, at the air-fuel ratio of 14.7 (stoichiometric air-fuel ratio), the rotation change distribution has a normal distribution, and the variation of the rotation change is relatively small. On the other hand, when the air-fuel ratio is shifted to the lean side, as in FIG 11b and 11c is shown, the dispersion of the rotational change is large and the amount of scattering is larger as the air-fuel ratio is shifted more toward the lean side. When the air-fuel ratio is shifted to the lean side, the rotational change that occurs at the position other than the normal distribution is measured in the small range of the rotational change.

If the air-fuel ratio is shifted to the lean side, is likely as an effect in each cylinder of the internal combustion engine a misfire occur and the rotation change at every burn will probably not appear on the ground the misfire. The rotation change distribution gives way from the normal distribution due to the misfire.

If an internal combustion engine is operated in an environment, wherein the combustion is likely to be bad if, for example unexpected misfires occur while the air-fuel ratio in the lean area is regulated, is the reliability the combustion state bad, which is determined on the basis combustion roughness calculated from the standard deviation, and the bad reliability is a problem.

It For example, a method may be used wherein the combustion pressure or the amount of combustion light in the combustion chamber is detected, to indirectly the combustion state in an internal combustion engine to capture, or a process, with the change in performance the rotation change is measured at an internal combustion engine to indirectly the combustion state to detect in the internal combustion engine.

There however, the combustion state is not necessarily related to the air-fuel ratio in a one-to-one relationship and the detection involves an error due to the detection procedure, It is difficult to understand the relationship between the air fuel ratio and to quantify the combustion state when the combustion state is more stable. In particular, when the combustion state of the rotation change is determined, the scatter of the relationship between the air-fuel ratio and the combustion state probably serious.

When the combustion state is to be detected while maintaining the air-fuel ratio in the lean state for a predetermined time at the state of large dispersion of the relationship between the combustion state and the air-fuel ratio, as in FIG JP 61-096150 A Accordingly, the driveability becomes poor due to the dispersion of the combustion state, and the bad driveability is a problem. A method in which the air-fuel ratio is once shifted to the lean side to some degree is not necessarily sufficient to determine whether or not the lean degree is the misfire limit. It is necessary to change the lean degree of the air-fuel ratio several times to repeatedly determine whether or not this lean degree corresponds to the misfire limit and to determine the optimum lean degree. In this method, it is necessary to control the air-fuel ratio in order to repeatedly shift the air-fuel ratio to the lean side near the misfire limit, and a lot of time and many control operations are required, and this is a problem of the method.

DE 44 07 167 A1 discloses a method for determining the torque transmitted by gas forces to the crankshaft of an internal combustion engine. Here, a torque Mgas is analyzed whereas no calculation of a combustion roughness as described in claim 1 is shown.

DE 195 40 675 C1 discloses a method for torque estimation by means of speed evaluation at the crankshaft of an internal combustion engine, wherein the elimination of noise components is considered here.

The The present invention has been made in view of the above problems made. The object of the invention is to provide a combustion state control device for an internal combustion engine, which is able to correct the combustion state to capture even if the internal combustion engine in an environment is operated, the combustion state probably bad is.

Around to solve the task becomes the torque change amount calculated at each combustion of the internal combustion engine, as in claim 1 described.

advantageous Embodiments of the invention are set forth in the dependent claims.

Even if according to the above Structure of the combustion state becomes badly irregular and the rotation change amount distribution deviates from the normal distribution, the combustion accuracy can be calculated without using the specific rotation change data or with a lower weighting of the same, at the disturbing effect of the bad combustion state are involved, since the only data included in a given furnished area statistically processed as the effective data. Therefore the disadvantage is avoided that the combustion state detected incorrectly becomes. Even if the internal combustion engine operated in an environment is, the combustion state is probably bad, As a result, the combustion state of the internal combustion engine detected correctly, and the combustion state is reliably detected.

In As used herein, the term "misfire" does not encompass just the case where the fuel in a combustion chamber of the internal combustion engine not ignited but also the case where fuel is ignited and the fire does not spread (semi-misfire).

embodiments The invention are illustrated in the drawing.

1 FIG. 12 is a structural diagram showing the outline of the air-fuel ratio control system of an internal combustion engine in accordance with the first embodiment of the present invention. FIG.

2 Fig. 10 is a flowchart of an air-fuel ratio control routine.

3 Fig. 10 is a flow chart of a Ne interrupt routine.

4 shows a time chart of a rotation change in each combustion.

5a and 5b show flowcharts of roughness calculation routines.

6 FIG. 12 is a graph showing the relationship between the combustion roughness, the fuel consumption and the air-fuel ratio. FIG.

7 FIG. 12 is a graph showing rotational change rotational data ΔNe when the air-fuel ratio is lean.

8th Fig. 12 is a graph showing the rotational change ΔNe statistical data when the air-fuel ratio is lean.

9 Fig. 10 is a graph showing the relationship between the combustion roughness and the air-fuel ratio.

10 shows a diagram of the relationship between the speed and the crank angle.

11a to c show diagrams of the statistical data of the rotation change.

12 Fig. 10 shows a flow chart of an improved Ne interrupt routine.

13 shows a time chart of the rotation change for each combustion.

14a to 14c Tne point time deviation diagrams show the required time.

15a and b show diagrams for setting the corrections K1 and K2.

16 FIG. 10 is a flow chart of another modified Ne interrupt routine. FIG.

17 shows a diagram for setting the correction value K3.

18 FIG. 12 is a flowchart of the modified roughness calculation routine. FIG.

19 FIG. 12 shows a flowchart of the roughness calculation routine that the 18 follows.

20 FIG. 12 is a timing chart showing the operation of the modified embodiment shown in FIG 18 and 19 is shown.

21 shows a diagram for setting up a sample inventory parameter n.

22 shows a map for setting the basic value n1 of the sample inventory parameter.

23 Fig. 10 is a diagram for setting the coefficient f (g) with the learning frequency as the parameter.

24 shows a map for setting the target roughness Rgt.

25 FIG. 12 is a flowchart showing another modified procedure for calculating the roughness. FIG.

26 FIG. 12 is a timing chart of the roughness calculation routine in another modification. FIG.

27 FIG. 12 is a flowchart showing the main routine of an air-fuel ratio control system in accordance with the second embodiment of the present invention. FIG.

28 Fig. 12 shows a flow chart of a NE interrupt routine.

29 FIG. 12 is a flowchart showing a target air-fuel ratio calculation routine. FIG.

30 FIG. 12 is a flowchart of a target air-fuel ratio calculation routine that the FIG 29 follows.

31 FIG. 10 is a flowchart showing the execution of the combustion state feedback routine. FIG.

32 FIG. 10 is a flowchart of a representative air-fuel ratio control routine. FIG.

33 shows a timing chart of the operation of the second embodiment.

34 shows a time chart of the rotation change for each combustion.

35 Fig. 10 is a graph showing the relationship between fuel consumption, combustion roughness and air-fuel ratio.

36 shows a map for setting the target roughness.

37 shows a diagram for setting the target air-fuel ratio correction.

And 38 shows a map for setting the rich change.

embodiments The present invention will be described below in detail with reference to various embodiments. At the following embodiments For example, the present invention is directed to an air-fuel ratio control system applied, which is used for a return regulation the air-fuel ratio of the air-fuel mixture to a desired value. In particular, the Combustion roughness detected as the combustion state of a Internal combustion engine and the air-fuel ratio is suitably controlled in the lean area in agreement the roughness. The structure and operation of the present invention are described in detail below.

First embodiment

In 1 has an engine 10 a six-cylinder internal combustion engine of the spark-ignited type. The sucked intake air from the air cleaner 11 passes through a suction line 12 and a throttle 13 through and is mixed with fuel passing through an injector 14 is injected, which is provided in the intake passage of each cylinder. Thereafter, the air-fuel mixture becomes each cylinder of the engine 10 fed. The exhaust gas emitted from each cylinder after combustion is released into the atmosphere via an exhaust manifold 15 and an exhaust pipe 16 ,

The suction line 12 is with an intake air temperature sensor 17 provided for detecting the temperature of the intake air and the air flow meter 18 for detecting the amount of intake air. The throttle 13 is with a throttle sensor 19 provided for detecting the opening angle of the throttle valve 13 , In addition, the exhaust pipe 16 with an air-fuel ratio sensor 20 for detecting the air-fuel ratio based on the oxygen concentration in the exhaust gas. A cylinder block is with a water temperature sensor 21 provided for detecting the temperature of the engine cooling water as another sensor used in this system. For example, a crankshaft with a crank angle sensor 22 for generating a Ne pulse signal at each timing of a predetermined crank angle (every 10 ° crank angle in the present embodiment).

An ECU 30 has a known microcomputer with a CPU 31 , a ROM 32 , a ram 33 and a backup RAM 34 on. The ECU 30 picks up the detected signals from various sensors, such as the intake air temperature sensor 17 , the air flow meter 18 , the throttle sensor 19 , the air-fuel ratio sensor 20 , the water temperature sensor 21 and the crank angle sensor 22 to the fuel injection amount of the injector 14 to determine and control the air fuel ratio in the desired manner. The CPU 31 successively executes various control routines according to the control program stored in the ROM 32 is stored. The backup RAM 34 is a memory for storing the content with the aid of the power supply from the vehicle battery, which is not shown in the drawing, and stores the basic map data for setting the target air-fuel ratio in accordance with, for example, the engine speed Ne and the intake air amount Qa.

When next The operation of the air-fuel ratio control system will be the same as that described above described structure described.

2 shows a flowchart of the by the CPU 31 performed air-fuel ratio control routine.

At the step 101 in 2 be the engine speed Ne, which from the detection result of the crank sensor 22 is calculated, and the intake air amount Qa, by the detection result of the air flow meter 18 is calculated, entered. At the following step 102 For example, the basic injection amount Pt is calculated on the basis of the input Ne and Qa using a map not shown in the figure.

After that, at the step 103 the cooling water temperature THW resulting from the detection result the water temperature sensor 21 is calculated, the intake air temperature TA, from the detection result of the intake air temperature sensor 17 is calculated, and the intake air amount change ΔQa (change of Qa during the interval of 720 ° crank angle) between adjacent combustion cycles of the same cylinders is input. At the step 104 the correction value K1 is calculated on the basis of THW, TA and ΔQa. The correction value K1 is a known fuel quantity correction for increasing / decreasing the control of the fuel during the cold operation or the transient operation.

At the step 105 the target air-fuel ratio λtg is calculated on the basis of the engine operating condition (Ne and Qa) of the time by the use of the retrieval map stored in the backup RAM 34 is stored. At the following step 106 That is, the air-fuel ratio correction value K2 is calculated based on the deviation of the air-fuel ratio using the actual or true air-fuel ratio λre detected by the air-fuel sensor 20 and the target air-fuel ratio λtg. Here, the air-fuel ratio correction value K2 is the known feedback correction calculated in accordance with the deviation of the air-fuel ratio.

Then at the step 107 the final injection amount Tau is calculated (Tau = Tp × K1 × K2) using the calculated basic injection amount TP and the correction values K1 and K2, and the routine is ended.

On the other hand, the CPU performs 31 a ne-interruption out in 3 is shown based on a NE pulse signal received from the crank angle sensor 22 every 20 degrees crank angle is supplied.

At the step 201 in 3 For example, the crank angle is calculated between the adjacent Ne pulse signals using the crank angle counter, which is not shown in the figure. At the following step 202 the required time of a given crank angle (20 ° crank angle) Tne is calculated and at step 203 For example, the rotational change ΔNe is calculated on the basis of the required time Tne for each combustion. Here at the six-cylinder engine, the required time Tne changes with a combustion cycle of 120 ° crank angle, as in 4 is shown, and the rotational change ΔNe is calculated from the difference between the maximum value Tmax and the minimum value Tmin. After that at the step 204 the average value of the ΔNeav of the rotational change ΔNe is calculated, and this routine is temporarily terminated.

Next, the calculation of the roughness will be described using the in 5a shown flow chart. The CPU 31 Runs this routine every time fuel is injected into each cylinder. Specifically, in this routine, the roughness (true roughness Rre) representing the true combustion state is calculated based on the standard deviation, and the true roughness Rre is calculated based on the standard deviation of the rotation variation of a given sample inventory parameter n.

As described above, the unexpected misfire is likely to occur when the air-fuel ratio is controlled in the lean region, and the rotational deviation ΔNe indicating the deviation from the normal distribution increases. 7 shows statistical data of the rotational deviation ΔNe during the lean control. In 7 the rotation change ΔNe is shown in the X-range. ΔNe is regarded as the effective data only when the rotational change ΔNe is equal to or larger than the average value ΔNeav, to eliminate the disturbing effect due to misfires. Here, the standard deviation is calculated using only the effective data, and the true roughness Rre is obtained.

• – Das Luftkraftstoffverhältnis wird in dem mageren Bereich geregelt;
• – Die Kühlwassertemperatur THW ist gleich oder höher als 80°C (Aufwärmzustand); und
• – Die Änderung der Motordrehzahl Ne in 180° Kurbelwinkel ist gleich oder kleiner als ein vorgegebener Wert (stetiger Betriebszustand).

In 5a is first at the step 301 determines whether the state for calculating the true roughness Rre is true or not. The state for execution includes, for example:

• - The air-fuel ratio is controlled in the lean range;
• - The cooling water temperature THW is equal to or higher than 80 ° C (warm-up condition); and
• - The change of the engine speed Ne in 180 ° crank angle is equal to or smaller than a predetermined value (steady state).

When at the step 301 yes, the routine moves to the step 302 and the rotational change ΔNe that occurs at the in 3 is calculated, and its mean value ΔNeav is read out. At the following step 303 It is determined whether or not the rotational change ΔNe is greater than the average value ΔNeav when ΔNe is greater than ΔNeav. The routine then proceeds to the following step 304 continued.

At the step 304 is the square of the difference between this value of the rotational change ΔNe (i) and of the average value ΔNeav is calculated by using the rotational variation of the ΔNe at the time when ΔNe is larger than ΔNeav, and the square value is accumulated. At the step 305 Then, the counter for counting the number of injections of each cylinder is incremented by one.

After that at the step 306 it is determined whether the fuel has been injected n times after the previous calculation of the roughness for each cylinder on the basis of the count value at the step 305 , If the result at the step 307 Yes, then the true roughness in Rre is calculated based on the standard deviation of the rotational change ΔNe of each cylinder. In detail, the true roughness Rre is calculated according to the following equation (1).

After that at the step 308 The target roughness Rtg is calculated by using a recovery map not shown in the figure, which is set in advance based on the engine speed Ne and the intake air amount Qa. In this case, a smaller target roughness Rtg is set when the speed is higher and the load is lower. On the other hand, a higher target roughness Rtg is established when the speed is lower and the load is higher.

As in 6 is shown, there is a correlation between the combustion roughness, the fuel consumption rate (fuel consumption) and the air-fuel ratio, and the target roughness Rtg is set in accordance with the minimum fuel for the best torque. In 6 , which covers the low leanness including the stoichiometric air-fuel ratio, the region A indicates the range where the combustion is stable and the change in the combustion roughness is not constant from the air-fuel ratio. On the other hand, the region B shows the region where the combustion is unstable due to the lean shift of the air-fuel ratio, and the roughness changes from the air-fuel ratio by the ratio of 1: 1. The target air-fuel ratio λtg during the lean control is set in the area B in 6 ,

After that at the step 309 For example, the deviation ΔR between the target roughness Rtg and the true roughness Rrt is calculated as ΔR = Rtg-Rre. At the following step 310 it is determined whether the roughness deviation ΔR is greater than a positive predetermined value (+ Kla) or not.

If ΔR> + Kla, then the target air-fuel ratio λTg is shifted to the lean side by a predetermined amount in the step 312 on the other hand, if ΔR <-Kla, then the target air-fuel ratio λtg is shifted to the rich side by a predetermined amount in the step 313 , The basic map data in the backup RAM 34 are renewed on the basis of the target air-fuel ratio λTg shifted to the lean side or rich side. Moreover, if -Kla ≦ ΔR ≦ + Kla, then this routine is brought to an end without changing the target air-fuel ratio λTg.

Short if the true roughness Rre is less than the target roughness Rtg, then the target air-fuel ratio λTg learns the lean side Value, and on the other hand, if the true roughness Rre is greater as the target roughness Rtg, then the target air-fuel ratio λTg learns the rich-side value. It will in this context, the deviation of the roughness for each Cylinder determines, when the map of the target air-fuel ratio λTg is common to all cylinders, The map data learns the average value of all cylinders. About that For example, each cylinder bank has a map at one V engine, where the map data is the average for each cylinder bank learn.

According to the present embodiment described above the following effects are foreseen.

Only the data in the previously established established range of rotational variation ΔNe for each combustion is considered to be effective and the true roughness Rre is calculated based on the standard deviation of the rotational variation ΔNe of the effective predetermined sample inventory parameter N. Specifically, only the rotational deviation ΔNe equal to or is greater than the average value ΔNeav of the rotation change, is considered effective, and the true roughness Rre is calculated. Even though the distribution of the rotational deviation ΔNe deviates from the normal distribution due to the misfires or the Similarly, when the air-fuel ratio is controlled in the lean range, therefore, the true roughness Rre is calculated without using the specific rotation-change data involved in the disturbing effect of the bad combustion state due to misfires or the like. Therefore, the disadvantage that the combustion state is erroneously detected is prevented. As a result, the combustion state of the engine becomes 10 detected correctly and the combustion state is detected reliably.

There the standard deviation is calculated for the rotational change ΔNe of a given sample inventory parameter N and the true roughness Rre is determined on the basis the standard deviation calculated as described above is, the combustion state is detected correctly in accordance with the dispersion of the rotation change.

There the combustion state (true roughness Rre) is detected appropriately as described above, is controllability the lean air-fuel ratio regulation which is activated in accordance with the combustion state. This means, because the air-fuel ratio and the combustion state (true roughness Rre) in correlation each other in the lean air-fuel ratio range, is the detection accuracy of the combustion state as above described improved, the lean target air-fuel ratio is learned using of the detected value to thereby improve the learning accuracy. About that In addition, the lean target air-fuel ratio learning acts for Realizing the air-fuel ratio regulation in accordance with the individual difference and change over time.

• (1) Nur die Drehänderungsdaten, die gleich oder kleiner als der Durchschnittswert ΔNeav sind, werden wirksam betrachtet und die wahre Rauigkeit Rre wird berechnet auf der Grundlage der Drehänderung ΔNe dieses wirksamen vorgegebenen Probenbestandsparameters, oder
• (2) Nur die Drehänderungsdaten, die in einem vorgegebenen Bereich mit der Mitte bei dem Durchschnittswert ΔNeav der Drehänderung enthalten sind, werden als wirksam betrachtet und die wahre Rauhigkeit wird berechnet auf der Grundlage der Drehänderung ΔNe des wirksamen vorgegebenen Probenbestandsparameters.

In the above embodiment, only the rotational change ΔNe larger than the average value is considered to be effective, and the true roughness Rre is calculated based on the standard deviation of the effective data. However, this procedure can be changed as follows. Examples:

• (1) Only the rotation change data equal to or smaller than the average value ΔNeav is considered to be effective and the true roughness Rre is calculated on the basis of the rotation variation ΔNe of this effective predetermined sample inventory parameter, or
• (2) Only the rotation change data included in a predetermined range centered on the average value ΔNeav of the rotation change is considered to be effective, and the true roughness is calculated based on the rotation variation ΔNe of the effective predetermined sample inventory parameter.

In detail, for (2) only the data contained in +/- σ, +/- 2σ or +/- 3σ is considered to be effective, where σ is the standard deviation of the rotational variation ΔNE. 8th shows the effective data range +/- σ, the effective data range +/- 2σ and the effective data range +/- 3σ. The true roughness Rre is calculated using the effective data.

Either in (1) as well as (2), the rotation change data used by the Normal distribution deviate due to misfires or the like, not used in a confusing way instead of the rotation modification data, which in the normal Distribution are included, and the true roughness Rre will be appropriate calculated. That is why the reliability of the true roughness Rre improves and the air-fuel ratio control with high accuracy is realized as a result.

In the embodiment of the above No. (2), since only the data included in +/- σ, +/- 2σ or +/- 3σ is considered effective, the spread range is limited in the correlative relationship between the combustion accuracy and the air-fuel ratio. 9 shows the limited range of the relationship between the combustion accuracy and the air-fuel ratio when the data is considered effective in the range of +/- σ, +/- 2σ or +/- 3σ.

Although the standard deviation is calculated to calculate the combustion roughness by means of statistical processing in the above embodiment, the mean square may be calculated as the statistical processing to calculate the combustion roughness, or the mean deviation may be calculated to calculate the combustion roughness , On the other hand, a means for calculating based on the combustion pressure of each combustion or a means for calculating based on the combustion light may be applied instead of the rotation change as the means for calculating the torque change at each combustion of the engine. In any case, only the torque change in the predetermined range set is considered to be effective, and the combustion roughness is calculated on the basis of the torque variation of the effective predetermined sample inventory parameter, and as a result, the combustion roughness (combustion condition).

Furthermore, the roughness calculation ( 5a ) alternatively be carried out as in 5b is shown. In this processing, the roughness is calculated using not only the data within the predetermined range but also other data outside the predetermined range while differentiating the weighting of the data. In particular, the data within the predetermined range is used with higher weighting, but the data outside the predetermined range is used with lower weighting. For example, the rotational change ΔNe larger than ΔNeave is regarded as data within the predetermined range.

As in 5b is shown, the roughness R is calculated as R1 and R2 in steps 304a and 304b in each case depending on the determination yes or no at the step 303 , The counts N1 and N2 are then counted up after the steps 304a and 304b , The sum (N1 + N2) of the counts N1 and N2 is compared with the predetermined number (N) of the injections at the step 306 , Then at the step 307a a weighting constant K determined based on the engine operating condition. Specifically, the weighting constant K is determined based on the engine speed and the throttle opening angle. For example, it is set to 0 in a normal engine operating condition, with the engine speed and throttle angle being generally in a proportional relationship, while being set to a larger value when the engine condition is an acceleration or deceleration condition that is outside normal Engine operating state is located.

After the step 307a becomes the true roughness Rre at the crotch 307b calculated as Rre = R1 / N1 + K × R2 / N2 using the weighting constant K. When the deceleration or acceleration of the motor occurs, the number of data within the predetermined range decreases. In this regard, however, the data outside the predetermined range with a higher weighting is used to calculate the true roughness Rre. Thus, the true roughness Rre can be accurately calculated.

First modification of the first embodiment

The modification includes the method of calculating the rotational variation ΔNe in the first embodiment (FIG. 3 ).

When the speed of the engine 10 changes from the minimum value to the maximum value, the rotational change ΔNe for each combustion is calculated on the basis of the change amount of the rotational speed. In the present embodiment, with the monitoring for the required time Tne continuing from 20 ° crank angle, the required time Tne is an inverse number of the rotational speed, the rotational change ΔNe for each combustion is calculated on the basis of the change of Tne that occurs when the required time Tne changes from the maximum value to the minimum value.

In detail, in a six-cylinder engine, the required time Tne changes with the period of 120 ° crank angle as a combustion cycle, as in FIG 13 and the required time Tne is measured six times during the 120 ° crank angle as T1, T2, T3, T4, T5 and T6. At this time, the maximum value Tmax (minimum speed) of the required time is detected using T1 to T3, and the minimum value Tmin (maximum speed) of the required time is detected using T4 to T6. Then, the rotational change ΔNe is calculated on the basis of the difference between the maximum value Tmax and the minimum value Tmin. The range L1 ranging from T1 to T3 and the range L2 ranging from T4 to T6 correspond to a predetermined range set to have the estimated peak point approximately in the middle.

If the maximum value or the minimum value of the required time Tne be measured without deviation from the center of the range L1 or of the area L2, the maximum value of the required time becomes Tne measured at T2 in agreement with the esteemed Peak point and the minimum value of the required time is measured at T5 in accordance with the esteemed Top point on similar Wise.

The However, speed is approaching approximately along the parabola the peak point and the behavior of near the peak point (maximum value) For example, the required time Tne observed speed will be divided into three patterns as described below.

A pattern with the tip at T3 (leading pattern), as in 14a is shown.

A pattern with the tip at T2, as in 14b is shown.

A pattern with the tip at T2 (lagging pattern), as in 14c is shown.

at the present embodiment T1 and T3 are corrected in accordance with the deviation from T2 including the estimated peak point, to get the peak.

Here is at the in 12 shown Ne interrupt routine, as with reference to 3 The counting is performed between Ne pulse signals with the interval of 20 ° crank angle by using the crank angle counter at the step 201 , At the following step 202 For example, the required time Tne of a predetermined crank angle (20 ° crank angle) is calculated and Tne coincides with the row of crank angles and is stored as the value of one of these T1 to T6.

After that, at the step 2030 determines whether this interruption corresponds to the detection timing of the maximum value Tmax or not. At the step 2040 It is determined whether or not this interruption corresponds to the detection timing of the minimum value Tmin.

If T1 to T3 of the cylinder are measured thereby to detect the maximum value Tmax, then a result is obtained at step 2030 and the routine goes to the step 2050 and the maximum value Tmax of the required time Tne is calculated using the following equation (2). On the other hand, if T4 to T6 of the cylinder are measured to detect the minimum value Tmin, then a result is obtained in step 2040 and the routine goes to the step 2060 and the minimum value Tmin of the required time Tne is calculated using the following equation (3). However, if a result of no is obtained in both steps 2030 and 2040 , the routine is brought to an end for the time being. Tmax = T2 + (| T3 - T1 | × K1 + K2) (2) Tmin = T5 - (| T6 - T4 | × K1 + K2) (3)

In the equations (2 and 3), a correction value is established by using the relationship described in, for example, FIG 15a is shown. According to 15a the correction value K1 increases with an increase of | T3 - T1 | or | T6 - T4 | and decreases with an increase in the engine speed Ne. K2 is set up using the relationship described in, for example, 15b is shown. According to 15b the correction value K2 decreases with an increase in the engine speed Ne.

Instead of calculating Tmax and Tmin using the above Equations (2) and (3), the correction expression including | T3 - T1 | or | T6 - T4 | and the engine speed Ne is determined as the parameter directly through a map search to thereby calculate Tmax and Tmin by adding or subtracting the correction term.

After that at the step 2070 For example, the rotational change ΔNe for each combustion is calculated by using the maximum value Tmax and the minimum value Tmin of the required time Tne calculated as described above, and this routine is terminated.

The Differences between the required times Tne, which is before and after the esteemed Peak point (T3 - T1 and T6 - T4) are calculated and the required time Tne (T2 and T5) at the estimated Top point is corrected in accordance with the difference. The bigger the Difference between the required times Tne is that before and after the esteemed Peak point (T3 - T1 and T6 - T4) the bigger it is the correction term at the estimated peak point (the second Expression of the right side in the above equations (2) and (3) for the correction of the required time Tne (T2 and T5). Furthermore the correction values K1 and K2 are set in accordance with the engine speed Ne. This allows the maximum value Tmax and the minimum value Tmin of the required time Tne is calculated accurately become. Even if the peak of required tne (speed) from the estimated Peak point deviates from leading or lagging, can result in that the combustion state (combustion roughness) are accurately detected without disturbing Effect of the deviation.

In particular, in the engine with external mixture formation, the air-fuel ratio is required with high accuracy as compared with the direct-injection engine, and the slight air-fuel ratio error may cause the disturbance to the controllability, such a severe demand tion can be accommodated in the device of the present invention.

There in the process of crank angle range (120 ° crank angle range), which is necessary is for a combustion divided into six subregions and the maximum value Tmax and the minimum value Tmin of the required time Tne is calculated are divided into three consecutive six times each Subareas, will set up a predetermined area including of the esteemed Peak point and the measurement of the required time tne in one given area simply performed.

There the standard deviation of the rotational change ΔNe of the given Sample inventory parameter n is calculated and the true roughness Rre is determined on the basis of the standard deviation, the Combustion state can be suitably detected in accordance with the dispersion the rotation change.

When next Another modification for calculating the rotational variation Δne of the first embodiment described in the present invention. In this modification is the difference between the estimated peak point and the leading time Tne calculates and the difference between the esteemed Peak point and the lagging required time Tne is calculated and the required time Tne of the estimated peak point is corrected in accordance with the ratio of these Differences of Tne.

16 Fig. 10 is a flowchart particularly showing the Ne interruption routine, wherein the required time Tne at the estimated peak point is corrected by applying this routine and the true peak value is calculated.

At the step 2030 in 16 It is determined whether or not this interruption coincides with the detection timing of the maximum value Tmax. At the step 2040 It is determined whether or not this interruption coincides with the detection time of the minimum value Tmin.

If T1 to T3 of this cylinder is measured to detect the maximum value Tmax, then a result is obtained at step 2040 and the routine goes to the step 2042 and the ratio of the absolute value of the difference between the required time T1 and T2 to the absolute value of the difference between the required time T2 and T3 (| T2-T1 |) / (| T3-T2 |) is calculated. In addition, at the following step 2050A the correction value K3 is calculated using the in 17 shown relationship on the basis of (| T2 - T1 |) (| T3 - T2 |) and the engine speed Ne. Next, the maximum value Tmax of the required time Tne is calculated using the following equation (4). Tmax = T2 + K3 (4)

According to the in 17 As shown, the correction value K3 is set to be larger when (T2-T1) is more different from (T3-T2) (when the value of the axis of abscissa is far from 1). Moreover, the correction value K3 is set to be smaller as the engine rotation speed Ne is higher. However, if (T2-T1) or (T3-T2) is 0, the maximum value Kmax is set as the correction value K3.

If T4 to T6 of this cylinder is measured to detect the minimum value Tmin, then a result is obtained in step 2040 and the routine proceeds to the step 2041 and the ratio of the absolute value of the difference between the required time T4 and T5 to the absolute value of the difference between the required time T5 and T6 (| T5 - T6 |) / (| T6 - T5 |) is calculated. In addition, at the following step 2060a the correction value K3 is calculated using the in 17 shown relationship on the basis of (| T5 - T4 |) / (| T6 - T5 |) and the engine speed Ne. Next, the minimum value Tmin of the required time Tne is calculated using the following equation (5). Tmin = T5 - K3 (5)

If however (T5 - T4) or (T6 - T5) is 0, the maximum value Kmax becomes the correction value K3 set up.

After that at the step 2070A For example, the rotational variation ΔNe for each combustion is calculated as ΔNe = Tmax-Tmin using the maximum value Tmax and the minimum value Tmin of the required time Tne calculated as described above, and the routine is terminated.

If a result no is obtained at both steps, the routine becomes provisionally as she is finished.

By this modification becomes the maximum value Tmax and the minimum value Tmin the required time Tne calculated exactly as in the first modification. Even though the peak time required Tne (speed) from the estimated Leading point deviates leading or lagging, can in consequence of the Combustion state (combustion roughness) are accurately detected without disturbing Effect of the deviation.

One another method can be used, for example, wherein the required time Tne is measured at 5 points, namely the estimated Top spot, the leading two points and two lagging Points, and the peak point is calculated using this Measurement results. However, it is required that the required Time Tne with shorter ones Interval is measured, for example, the crank angle interval for the Measurement at 10 ° crank angle narrows when the number of tne measurement data is increased.

There the rotational state for Each combustion is determined by monitoring, can the combustion roughness calculated on the basis of the maximum value of the speed. In detail, the maximum value of the speed corresponds to the minimum value Tmin of the required time Tne and the true roughness Rre is calculated based on the standard deviation of Tmin.

Second modification of the first embodiment

The second modification involves improving the roughness calculation of the first embodiment ( 5 ). More specifically, in this routine, the roughness representing the true combustion state (true roughness Rre) is calculated based on the standard deviation, and the true roughness Rre is calculated based on the standard deviation of the rotation variation of the sample inventory parameter n.

When at the step 301 if a result is obtained, then the routine goes to the step 3020 and it is determined whether the sample inventory parameter n is set up or not. If the sample inventory parameter n has not yet been set up, then the sample inventory parameter becomes steps 3030 to 3110 set up.

In detail, at the step 3030 With reference to the previous value ΔRp of the deviation between the target roughness Rtg and the true roughness Rre, determines whether or not the absolute value of the preceding ΔRp is greater than the predetermined value Kr. If | dRp | > Kr holds, then the routine moves to the step 3040 and the sample inventory parameter n is established based on the roughness deviation ΔR. In principle, the sample inventory parameter is set up to be smaller when | ΔR | is larger. As in detail in 21 is shown when | ΔR | <R1, then n = n1, if | ΔR | = R1 to R2, then n = n2 and if | ΔR | > R2 holds, then n = n3.

On the other hand, if | ΔRp | ≦ Kr, the routine goes to step 3050 with the engine operating information such as the engine speed ne and the intake air amount Qa being input. At the following step 3060 the base value n1 of the sample inventory parameter is calculated using the previously prepared recovery map displayed in 22 is shown, based on the engine speed ne and the intake air amount Qa of the time. According to 22 The base value n1 is set to be smaller for a higher speed and the higher load range. On the other hand, the base value n1 is set to be smaller for lower speed and lower load.

In addition, at the step 3070 Determines if the battery change was directly preceding or the lean air-fuel ratio was not learned. When at the step 3070 a result is obtained yes, then at the step 3080 the base value n1 is set up as the sample stock parameter n, and the number of the learning time g is set to 0 at the step 3090 ,

If a result is no at the step 3070 , then the number of study times g at the step 3100 and the sample inventory parameter n is calculated in accordance with the number of learning times g at the step 3110 , In this case, a coefficient f (g) including the number of learning times g as the parameter according to the in 23 calculated relationship calculated. According to 23 the coefficient f (g) is roughly calculated so that the sample inventory parameter n increases with an increase in the number of samples Study times g.

After setting up the sample inventory parameter n, step 306 to 309 in 19 ( 5 ). At the step 308 the target roughness Rtg is calculated using the previously established recovery map shown in FIG 24 is shown, based on the engine speed ne and the intake air amount Qa. According to 24 For example, the target roughness Rtg is set to be smaller for a higher speed and lower load. On the other hand, the target roughness Rtg is set to be larger for a lower speed and higher load.

After that, at the step 3160 determines whether or not the true air-fuel ratio λRe converges to the target air-fuel ratio λTg. When the deviation between the true air-fuel ratio λRe and the target air-fuel ratio λTg is equal to or smaller than a predetermined value, and the air-fuel ratio is determined to be convergent (if at step 3160 a result is obtained yes), then the target air-fuel ratio λTg becomes at steps 310 to 313 learned.

Finally, at the step 3210 the number of learning times g increases by 1 and the routine is then temporarily ended. The data of the number of learning times are stored and stored in the backup RAM 34 held.

20 FIG. 12 is a flowchart showing the operation of the modified example and the process as an example in which the air-fuel ratio feedback control is transferred from the stoichiometric control to the lean control. FIG.

Before the time t1, the in 20 11, the closed-loop air-fuel ratio is controlled in the stoichiometric range and the true air-fuel ratio λRe shown by a solid line converges on the target air-fuel ratio λTg shown by a broken line. The true roughness Rre shown by a solid line converges to the target roughness Rtg shown by a broken line.

at At time T1, the target air-fuel ratio λTg of the stoichiometric Air-fuel ratio switched to a certain value in the lean area. Furthermore At time T1, the target roughness Rtg is changed from a relatively large one Value at the same time as switching to the lean air-fuel ratio control.

at At time T2 after time t1, the true air-fuel mixture λRe converges to the target air-fuel ratio λTg. There the true roughness Rre is smaller than the nominal roughness Rtg and the deviation ΔR Roughness (= Rtg - Rre) equal to or greater than is the predetermined value, the target air-fuel ratio λTg is changed to the lean side.

To the change the desired air-fuel ratio λTg at the At time T3, the true air-fuel ratio λRe converges again approximately to the target air-fuel ratio λTg. Because beyond that while the true roughness Rre is smaller than the nominal roughness Rtg and the deviation ΔR (= Rtg - Rre) equal to or greater than is a predetermined value, the target air-fuel ratio λTg is changed back to the lean side. After that the true roughness Rre converges to the set roughness Rtg and the true air-fuel ratio λRe converges to the target air-fuel ratio λTg.

At times T2 and T3, the target air-fuel ratio λTg learns the lean-side value based on the roughness deviation ΔR and the basic map data of the backup RAM 34 be renewed.

On the other hand, in the state that the true roughness Rre approximately converges to the target roughness Rtg, the sample inventory parameter n is set to be relatively larger. On the other hand, since the roughness deviation ΔR (= Rtg-Rre) increases, immediately after the switching of the target air-fuel ratio λTg, the sample inventory parameter n is changed to a relatively small "n3" and "n2" according to the in 21 shown relationship in the time period of Ta and Tb.

Therefore becomes immediately after switching the air-fuel ratio to the lean side of the sample inventory parameters n suppressed a small value, so that the response of the true roughness Rre is considered important. Thereafter, the sample inventory parameter becomes n increased, so that the convergence of true roughness is considered important becomes.

• (a) da der Probebestandsparameter n geeignet geändert wird in Übereinstimmung mit dem Motorbetriebszustand bei dem vorliegenden Gerät zum Berechnen der wahren Rauhigkeit Rre auf der Grundlage der Standardabweichung, werden das Ansprechverhalten und die Konvergenz der wahren Rauhigkeit Rre beliebig eingestellt. In Folge dessen kann der Verbrennungszustand korrekt erfasst werden, selbst wenn der Motorbetriebszustand sich auf verschiedene Weise ändert.
• (b) da der Probebestandsparameter n variabel eingerichtet ist in Übereinstimmung mit verschiedenen Faktoren, wie beispielsweise der Abweichung ΔR der Rauhigkeit, dem Motorbetriebszustand, der Batterieinitialisierung oder dem Lernvorgang des mageren Soll-Luftkraftstoffverhältnisses und der Anzahl der Lernzeiten g, kann der Verbrennungszustand korrekt erfasst werden mit einer bevorzugten Auswahl des Ansprechverhaltens oder der Konvergenz in Übereinstimmung mit dem Erfordernis der Zeit.
• c) da der Verbrennungszustand (wahre Rauhigkeit Rre) korrekt erfasst werden kann, wie vorstehend beschrieben ist, ist die Regelbarkeit der mageren Luftkraftstoffverhältnisregelung verbessert, die aktiviert wird in Übereinstimmung mit dem Verbrennungszustand. Da im Detail das Luftkraftstoffverhältnis eine Korrelation zu dem Verbrennungszustand (wahre Rauhigkeit Rre) in dem mageren Luftkraftstoffverhältnisbereich hat, bringt die Verbesserung der Erfassungsgenauigkeit des Verbrennungszustands, die vorstehend beschrieben ist, die Verbesserung der Lerngenauigkeit mit sich, da der erfasste Wert verwendet wird zum Lernen des mageren Soll-Luftkraftstoffverhältnisses. Darüber hinaus bringt das Lernen des mageren Soll-Luftkraftstoffverhältnisses die Verwirklichung der Soll-Luftkraftstoffverhältnisregelung in Übereinstimmung mit der individuellen Differenz und der Änderung über der Zeit mit sich.

According to the second modification, the following effects are provided.

• (a) Since the sample inventory parameter n is suitably changed in accordance with the engine operating condition in the present roughness calculation apparatus Rre based on the standard deviation, the response and the true roughness convergence Rre are arbitrarily set. As a result, the combustion state can be correctly detected even if the engine operating state changes in various ways.
• (b) since the sample inventory parameter n is variably set in accordance with various factors such as the roughness deviation ΔR, the engine operating condition, the battery initialization or lean target air-fuel ratio learning and the number of the learning times g, the combustion condition can be correctly detected with a preferred choice of responsiveness or convergence in accordance with the requirement of time.
• c) since the combustion state (true roughness Rre) can be detected correctly as described above, the controllability of the lean air-fuel ratio control is improved, which is activated in accordance with the combustion state. In detail, since the air-fuel ratio has a correlation with the combustion state (true roughness Rre) in the lean air-fuel ratio range, the improvement in the detection accuracy of the combustion state described above brings about the improvement of the learning accuracy since the detected value is used for learning the lean desired air-fuel ratio. Moreover, learning the lean target air-fuel ratio involves realizing the target air-fuel ratio control in accordance with the individual difference and the change with time.

A another modification for calculating the accuracy will be below described. In this modification, the standard deviation is calculated in reference value units using a certain sample inventory parameter as the reference value. The calculated results will be successive accumulated and the average value of the accumulated value used to calculate the combustion roughness.

25 FIG. 10 is a flowchart showing the true roughness calculation sequence in the present embodiment. FIG. This routine will be at the step 307 who in 19 shown is applied. At the in 25 In the routine shown, the base value Kc of the sample inventory parameter for calculating the roughness is set in the relatively small stock parameter unit, and the standard deviation of the sample stock parameter n of the time is calculated in Kc units.

At the step 3071 the counter k is incremented and at the following step 3072 It is determined whether the counter k reaches the reference value Kc or not. When at the step 3072 if a result is obtained, then the routine goes to the step 3073 and the roughness R1 of the small inventory parameter unit is calculated according to the following equation (6).

After that at the step 3074 the counter k is cleared to "0". At the following step 3075 the roughness r1 of the small parameter units is accumulated and the accumulated value is considered R2 (R2 = R2 + R1).

At the step 3076 the sample inventory parameter n is taken over time and at the following step 3077 it is determined whether the fuel injection is performed in accordance with the sample inventory parameter n or not after the calculation of the previous true roughness Rre. If a result is yes, then the routine proceeds to step 3078 continued. In the following embodiment, it is assumed that the sample inventory parameter n is larger than the reference value Kc.

At the step 3078 The true roughness Rre is calculated on the basis of the accumulated value of the calculated roughness, the sample inventory parameter n, and the reference value Kc according to the following equation.

After that at the step 3079 becomes the accumulated roughness 52 cleared to "0" and the routine is ended. In the 25 however, the routine shown can be applied only when the sample inventory parameter n is large.

26 shows a timing diagram of the actual operation, which in the in 25 shown routine is performed. In 26 For example, the roughness R1 of the small inventory parameter units is accumulated each time the counter k is counted by kc, and the accumulated value R2 of the roughness is calculated. When the number of injection of each cylinder reaches the sample inventory parameter n at time T10, the true roughness Rre is calculated according to the above-described equation at time T11 thereafter.

According to the second Variation, the roughness can be calculated in small inventory parameter units. That's why it's possible to perceive the trend of roughness on half of the calculation, if the sample inventory parameter is large and a big time is required for the calculation of roughness.

The first embodiment can also be performed as described below in addition to the above modifications.

Immediately after the calculation of the true roughness Rre, the true roughness Rre is subjected to a smoothing calculation. In detail, the smoothing rate Sn is set in advance, and the smoothing value is calculated according to the following equation. Smoothing value = previous smoothing value + (Rre - previous smoothing value) × Sm.

As an example, the reference value of the smoothing rate Sm is set to a 64th. At the in 19 shown steps 309 to 313 the deviation between the true roughness smoothing value Rre and the target roughness Rtg is calculated, and the learning of the target air-fuel ratio λTg is performed in accordance with the deviation.

there becomes the smoothing rate Sm (degree of smoothing) preferably changed in accordance with the engine operating condition, for example, the smoothing rate Sm big decorated and the degree of smoothing the true roughness Rre is set small when the response true roughness is considered important, for example in the transitional operation. On the other hand, the smoothing rate becomes Sm set small and the degree of smoothing the true roughness is made big, if the convergence of the true roughness Rre considered important is in steady operation. As a result, the combustion state becomes detected correctly, even if the engine operating state different changed and the controllability of the air-fuel ratio is improved.

at the first embodiment the standard deviation is calculated when the combustion roughness is calculated by the statistical processing. Instead however, the mean square may be calculated or the middle one Deviation as the statistical processing for calculating the combustion roughness. A means can be used, with the torque change calculated by the combustion pressure at each combustion, or a means, wherein the torque change by the combustion light can be used as another means to Calculate the torque change as the torque change.

Second embodiment

An air-fuel ratio control system in accordance with a second embodiment employs a 27 shown main routine.

When an ignition switch not shown in the figure is turned on and a motor 10 is started, the in 27 The main routine shown is activated and the initialization becomes at step 2110 carried out. At the step 2120 will be the engine speed n3 calculated on the basis of that from the crank angle sensor 22 supplied input result and the intake air amount Qa, which is calculated on the basis of the air flow meter 18 supplied detection result, and at the following step 2130 For example, the basic injection amount Tp is calculated on the basis of Ne and Qa using a map not shown in the figure.

After that, at the step 2140 entered the following data: The cooling water temperature THW, which is calculated on the basis of that of the water temperature sensor 21 supplied intake result, the intake air temperature Ta, which is calculated based on the of the intake air temperature sensor 17 supplied intake result, and the intake air amount change ΔQa (Qa change between 720 ° crank angle) of the consecutive combustion cycles of each cylinder. At the step 2150 The correction value K1 is calculated on the basis of the inputted THW, TH and ΔQa. The correction value K1 is a known fuel quantity correction for increasing or decreasing the fuel supply during the cold operation or the transient operation.

Next shows 28 a flowchart of the routine that is intermittently activated every 20 ° crank angle on the basis of the Ne-pulse signal from the crank angle sensor 22 is supplied. This routine is preferably applied to the main routine in the ne-interrupt routine.

At the step 2201 The crank angle counter, not shown in the figure, counts between adjacent Ne pulse signals with 20 ° crank angle interval. At the following step 2202 the required time Tne of the predetermined crank angle (20 ° crank angle) is calculated and at step 2203 For example, the rotational change ΔNe for each combustion is calculated on the basis of the required time Tne. Here at the six-cylinder engine, the required time Tne changes with a combustion cycle of 120 ° crank angle, as in 34 is shown, and the rotational change ΔNe is calculated from the difference between the maximum value Tmax and the minimum value Tmin. The data ΔNe is stored in the RAM 33 stored for each cylinder.

After that, at the step 2204 determines whether the combustion state detection is allowed or not. The combustion state detection condition includes, for example, the warm-up state and the steady-state operation, and in detail, the state that the cooling water temperature THW is equal to or higher than 80 ° C and the state that the change of the engine rotation speed Ne at 180 ° crank angle is equal to or smaller than a predetermined value ,

If the combustion state detection is allowable, then the routine proceeds to step 2205 continued. At the step 2205 the true roughness Rre is calculated as the parameter representing the true combustion state of time based on the standard deviation using the calculated rotational change ΔNe for each cylinder. In detail, the true roughness Rre is calculated by using the rotational variation of each cylinder, which is obtained this time as ΔNe (i), and the average value as ΔNeav and the sample inventory parameter as n in the same manner as in the first embodiment.

After that, at the step 2206 determines whether the current timing of the injection timing of a cylinder corresponds or not, and if the current time does not correspond to the injection timing of a cylinder, then the routine is temporarily terminated. On the other hand, if the current time corresponds to the injection timing of a cylinder, then that is determined by the air fuel sensor 20 detected air-fuel ratio (true air-fuel ratio λRe) at the step 2207 entered, and at the following step 2300 the desired air-fuel ratio λTg is calculated. At the step 2300 will be in 29 and 30 applied routines shown.

The air-fuel ratio correction value K2 is calculated on the basis of the deviation between the true air-fuel ratio λRe obtained at the step 2207 is entered, and at the step 2300 calculated target air-fuel ratio λTg. Here, the air-fuel ratio correction value K2 is a known feedback correction value calculated in accordance with the air-fuel ratio deviation.

Finally, at the step 2209 the final injection amount Tau is calculated (Tau = Tp × K1 × K2) using the calculated basic injection amount Tp and corrections K1 and K2, and the routine is ended.

Here, the calculation of the target air-fuel ratio λTg at the in 28 shown step 2300 describe with reference to flow charts, which in 29 and 30 are shown.

First, at the in 29 shown step 2301 determines whether the lean tax execution condition is true or not. The lean control execution condition includes, for example, that the cooling water temperature THW is equal to or higher than 60 ° C, no high speed and no high load state, and no idle state. If the lean control execution condition is not true, then the routine goes to step 2302 and the stoichiometric control is carried out with the target air-fuel ratio λTg of λ = 1.

If the lean control execution control is true, then the routine goes to step 2303 , and it is determined whether or not the combustion state detection is permitted and a predetermined time Tx has elapsed from allowing the combustion state detection. A predetermined time Tx is the steady state stabilization waiting time from the start of combustion state detection, and Tx is set to two seconds in the present embodiment. If a result is yes, then the execution of the combustion state feedback control is permitted. The combustion state is controlled in the target state so that the true roughness Rre is made equal to the target roughness Rtg, and the combustion state feedback execution flag is turned on at the step 2304 ,

At the step 2305 It is determined whether the combustion state feedback execution flag is turned on or not. If the combustion state feedback execution flag is turned off, then the processing in the steps becomes 2306 to 2308 executed. If the combustion state feedback is not executed in detail, at step 2306 entered various engine operating conditions such as Ne, Qa and Taw. At the following step 2307 For example, the basic air-fuel ratio λbse is calculated based on Ne and Qa of the time using the in the backup RAM 34 stored desired air-fuel ratio map. At the step 2308 is the value obtained by the correction of the calculated basic target air-fuel ratio λbse by a correction value KTHW for the water temperature as the target air-fuel ratio λTg, and the routine then returns to the in 28 shown back routine.

On the other hand, if a result is obtained yes at the step 2305 (If the combustion state feedback execution flag is turned on), the combustion state feedback at step 2320 executed in 30 is shown. At the next step 2330 For example, the representative air-fuel ratio λRep for representing the air-fuel ratio in the combustion state feedback is calculated.

Here, there is a correlation between the specific fuel consumption (fuel consumption), the combustion roughness, and the air-fuel ratio among each other, as in 35 is shown. In a range A, up to the weak lean region including the stoichiometric air-fuel ratio in 35 is sufficient, the combustion is stable, the combustion roughness is small and the change in the combustion roughness is not constant with the air-fuel ratio. On the other hand, the combustion becomes unstable when the air-fuel ratio is shifted to the lean side in the area B in FIG 35 and the roughness increases approximately 1: 1 with the increase in the lean shift of the air-fuel ratio. Here, the combustion state feedback is performed in a region B, and the relationship between the combustion roughness and the air-fuel ratio is quantified in the combustion state feedback. Then, the air-fuel ratio is corrected using the quantified relationship.

Next, the execution sequence of the combustion state feedback and the calculation sequence of the representative air-fuel ratio λRep will be described with reference to FIG 31 and 32 ,

In 31 becomes at the step 2321 the target roughness Rtg is calculated from the engine speed Ne and the intake air amount Qa of time. For example, the target roughness Rtg is set in the driveability permissible range near the misfire limit of the engine 10 using the in 36 shown map. As in 36 is shown, the target roughness Rtg is set to be smaller when the rotational speed Ne is higher and the load Qa is lower (higher Ne and smaller Qa). On the other hand, the target roughness Rtg is set to be larger as the revolving speed Ne is lower and the load Qa is higher.

At the step 2322 is the deviation ΔR (= Rtg - Rre) between that at step 2321 calculated target roughness Rtg and the step 2205 calculated true roughness Rre, who in 28 shown is calculated. At the following step 2323 the target air-fuel ratio correction value F is calculated on the basis of the roughness deviation ΔR using the in 37 shown relationship. As in 37 is shown, if ΔR> 0 (Rtg> Rre), then a positive target air-fuel ratio correction value F is set in accordance with ΔR. On the other hand, if ΔR <0, then a target negative air-fuel ratio correction value F is set in accordance with ΔR. The target air-fuel ratio correction f is monitored or limited when the absolute value of the roughness deviation ΔR exceeds a predetermined value.

After that at the step 2324 That is, a target air-fuel ratio correction value f is added to the previous value of the target air-fuel ratio λTg, and the added value is regarded as the new target air-fuel ratio λTg. If ΔR> 0 (Rtg> Rre), then the target air-fuel ratio λTg is corrected to the lean side. On the other hand, if ΔR> 0, then the target air-fuel ratio λTg is corrected to the rich side. By correcting the target air-fuel ratio λTg as described above, the combustion state is maintained in the desired state.

On the other hand, at the step 2331 in 32 determines whether or not a predetermined time Ty has elapsed from the start of combustion state feedback. A predetermined time ty is a time to wait for the stabilization of the true roughness Rre after the start of combustion state feedback, and Ty is 0.5 seconds in the present embodiment. If a result is yes at the step 2331 is received, then the routine proceeds to the step 2332 , the accumulated value Tλ of the true air-fuel ratio λRe (detected value of the air-fuel sensor 20 ) is calculated (Tλ = Tλ + λRe).

After that at the step 2333 It is determined whether or not a predetermined time Tz has elapsed from the start of the accumulation of the true air-fuel ratio λRe and the accumulation of a predetermined number of the samples. A predetermined time Tz is the sample time of the true air-fuel ratio λRe, and Tz is 10 seconds in the present embodiment. When at the step 2333 a result is obtained yes, then the average value of the accumulated true air-fuel ratio value Tλ is calculated at the step 2334 and this value is set as the representative air-fuel ratio λRep, and the sequence then returns to the in 30 shown back routine. Specifically, the accumulated value Tλ is divided by the number of injections Nc in the time period Tz and the value is treated as λrep (λRep = Tλ / Nc).

There the reliability of the representative Air fuel ratio λRep of the Probation (average time) depends on a given Time Tz be set variably in accordance with the desired Accuracy of the air-fuel ratio control.

Back to 30 after the combustion state feedback and the representative air-fuel ratio are calculated (step 2320 and 2330 ), becomes at the step 2309 determines whether the representative air-fuel ratio λRep is fully calculated or not. In the above routine, which in 32 is shown when the predetermined time Tz has not elapsed since the start of the accumulation of the true air-fuel ratio λRe (when in step 2333 no result is obtained no), then a result is no at the step 2309 determined.

In addition, when the representative air-fuel ratio λRep is calculated after the predetermined time Tz elapses from the start of the true air-fuel ratio accumulation λRep, the routine goes to step 2310 continued. At the step 2310 For example, the rich change ΔRech is calculated from the engine speed Ne and the intake air amount Qa of the time using the in 38 shown map. As in 38 is shown, the rich change ΔRech is set to be larger when the rotational speed Ne is higher and the load Qa is lower. On the other hand, the rich change ΔRich is set to be smaller when the revolving speed Ne is lower and the load Qa is higher.

At the following step 2311 the rich change ΔRech is subtracted from the representative air-fuel ratio λRep, and the obtained value is regarded as the target ratio λTg (= λRep-ΔRech).

At the following step 2312 becomes the target air-fuel ratio map in the backup RAM 34 renewed (learned) using the target air-fuel ratio λTg of the time as the basic target air-fuel ratio λbse. That is, the previous map data is overwritten with the calculated target air-fuel ratio λTg to coincide with the engine operating area of time. Finally at the step 2313 the combustion state feedback flag is turned off and this routine is ended.

When the target air-fuel ratio map for all cylinders is common, the Kenn field data learned using the average value of all cylinders. Moreover, for example, if each cylinder bank has an individual map as in a V-type engine, the map data is learned using the average value of each cylinder bank.

Next, the actual operation of the second embodiment will be described with reference to FIG 33 Shown timing diagram described. In 33 The air-fuel ratio feedback is regulated in a lean air-fuel ratio as the basic operation. For example, before time T11, the target air-fuel ratio λTg is set in a lean range based on the engine operating condition, and the true air-fuel ratio λRe changes in accordance with the target air-fuel ratio.

Of the Time T10 indicates the time when the combustion condition detection permissible and the true roughness Rre is calculated after time T10. At the time T11, when the predetermined time Tx since the time T10 elapses, the combustion state feedback execution flag is turned on. The given time Tx is the time to wait, if the true roughness Rre is calculated appropriately.

After the time T11, the target roughness Rtg becomes close to the misfire limit of the engine 10 and the combustion state feedback is performed so that the true roughness Rre converges to the target roughness Rtg. In other words, the target air-fuel ratio λTg is appropriately corrected in accordance with the roughness deviation ΔR, and the air-fuel ratio is feedback-controlled using the target air-fuel ratio λTg.

To the Time T12 when the predetermined time Ty since the time T11 the accumulation of the true air-fuel ratio λRe is started. At time T13, when the given time Tz has passed the start of the accumulation of the true air - fuel ratio λRe becomes the accumulated value Tλ of the true air-fuel ratio averaged to the most representative Air-fuel ratio λRep, the fat change ΔRech becomes subtracted from the representative Air fuel ratio λRep to the To calculate desired air-fuel ratio λTg, and the air-fuel ratio feedback is controlled using the desired air-fuel ratio λTg.

Furthermore At time T13, the previous basic target air-fuel ratio λbse is renewed Use of the desired air-fuel ratio λTg of the time (= λRep - ΔRech), and the combustion state feedback execution flag will be switched off.

The combustion state feedback control is performed at a point L1 near the misfire limit in FIG 35 and the air-fuel ratio at the point L1 is calculated as the representative air-fuel ratio λRep. The target air-fuel ratio λTg is learned at the point L2, and the representative air-fuel ratio λRep is shifted to the rich side by ΔRech. At the time, the air-fuel ratio at the point L1 corresponds to the minimum fuel for the best torque, and the target air-fuel ratio λTg is learned as the minimum fuel for the best torque.

According to the above described second embodiment the following effects are foreseen.

There the combustion state feedback control accomplished becomes in the lean air-fuel ratio range in the learning processing of the target air-fuel ratio λTg is the Combustion state fixed at a given state without Fluctuation in the period of the regulation. That's why the problem is solved that the air-fuel ratio unexpectedly exceeds the misfire limit and the combustion state as a result of the tolerance of the Driveability deviates. The engine is operated in a stable manner, since the target roughness Rtg is set in the tolerance of drivability near the misfire border, when the combustion state feedback control accomplished and poor drivability is prevented.

Since the ratio of the combustion roughness of the air-fuel ratio is about 1: 1 in the lean air-fuel ratio range, the representative air-fuel mixture λRep is obtained in a predetermined period of time when the combustion state feedback control is carried out in the lean-in region and the relationship between the combustion roughness (combustion state ) and the air-fuel ratio is quantified correctly based on the representative tative value. At this time, since the representative air-fuel ratio λRep required for the quantification has to be measured only once, the relationship between the combustion roughness and the air-fuel ratio can be quickly quantified, and as a result, the control time can be shortened considerably. In particular, when the combustion roughness is detected in accordance with the rotational change ΔNe, although the dispersion of the relationship between the combustion roughness and the air-fuel ratio is considerable in the conventional apparatus, such a problem is solved in the present embodiment, which belongs to the conventional apparatus, and excellent air-fuel ratio control can be realized.

Since the representative air-fuel ratio λRep is calculated on the basis of the air-fuel sensor 20 detected value (true air-fuel ratio λRe), the relationship between the roughness and the air-fuel ratio is quantified in consideration of the output error of the air-fuel sensor 20 , Therefore, the air-fuel ratio control is executed in consideration of the output error of the air-fuel sensor 20 , The relationship between the combustion state (combustion roughness) and the air-fuel ratio is suitably quantified, and the air-fuel ratio control is performed according to the relationship as described above, as a result of which the control accuracy is improved.

The representative air-fuel ratio λRep is shifted to the rich side by a predetermined amount to correct the target air-fuel ratio λTg, and the basic data of the target air-fuel ratio λTg are learned on the basis of the corrected result. Therefore, the air-fuel ratio control can be realized with high accuracy while quantifying the relationship between the combustion accuracy (combustion state) and the air-fuel ratio, and the output error of the air-fuel sensor 20 is taken into account. Moreover, the air-fuel ratio control is executed at the minimum torque for the best torque, and the air-fuel ratio control is desirably performed.

There the accumulated value tλ of the Air fuel ratio λRe obtained becomes in a predetermined time period of the combustion state feedback control and the average value of the accumulated value tλ is considered the most representative Air fuel ratio λRep, can the dispersion of the air-fuel ratio with respect to Combustion state can be reduced. As a result, increases the reliability of the representative Air fuel ratio λRep and the Control accuracy of the air-fuel ratio is further improved.

The second embodiment can do the following be modified. In detail, the air-fuel ratio correction value k2 changed be based on the representative air-fuel ratio λRep instead the learning of the desired air-fuel ratio λTg. On the other hand, the air-fuel ratio learning value becomes established on the basis of the representative air-fuel ratio λRep, the Fuel injection amount is corrected according to the air-fuel ratio learning value and the air-fuel ratio feedback control is running.

Although the representative air-fuel ratio λRep is calculated based on the average value of the true air-fuel ratio λRe in a predetermined sampling period, the representative air-fuel ratio λRep may be calculated instead based on the true-air-fuel-ratio smoothing value λRe in a predetermined sampling period. In detail, the calculation (steps 2332 to 2334 in 32 ) for accumulating and determining the true air-fuel ratio λRe replaced by the smoothing calculation. At this time, the true air-fuel ratio λRe is subjected to the smoothing calculation in a predetermined period of time when the combustion state feedback control is executed, and the smoothed value is regarded as the representative air-fuel ratio λRep. Since the load of the control software is reduced by employing the smoothing calculation, this embodiment is effective when the averaging time is established.

at the above embodiment the combustion roughness can be calculated using other statistical processing, such as a median Square or a mean deviation. Unlike the mean for calculating the combustion state by the rotation change, As described above, the means for calculating the Combustion pressure of each combustion or a means for calculating of the combustion light are used as the means for detecting the combustion state (roughness).

In particular, when calculating the rotational change .DELTA.NE (in 28 shown step 2203 ) in the second embodiment, the method described with reference to FIG 12 or 16 , which is described as the modified example of the first embodiment. Moreover, in the calculation of the roughness Rre in the second embodiment (which is described in FIG 28 shown step 2205 ) the method used with reference to 5 . 18 . 19 or 25 is described.

In the engine system according to the invention, the maximum value and minimum value of a rotational speed (Ne) are detected for each combustion of the engine ( 10 ). A rotation change (ΔNe) is calculated ( 203 ) based on a difference between the detected speed values. In addition, a statistical processing of the rotation change of a predetermined sample inventory parameter is carried out and a combustion roughness (ΔR) is calculated ( 307 ) as a parameter indicating the combustion state on the basis of the statistical processing result. In detail, only the data in a predetermined set range of rotational variation is considered to be effective for each combustion, and the true roughness is calculated based on the standard deviation of the rotational variation of the effective predetermined sample inventory parameter. For example, only the rotation change greater than an average value of the rotation change is considered to be effective for calculating the true roughness.

Claims (34)

A combustion state controller for an internal combustion engine, comprising: a torque change calculator ( 203 ) for calculating a torque change of an internal combustion engine ( 10 ) for each combustion; and a roughness calculator ( 303 to 307 ) for calculating a combustion roughness serving as a parameter for displaying the combustion state by statistically processing the torque variation of an effective predetermined sample inventory parameter under a condition that only data in a predetermined established range of the calculated torque variation is considered to be effective ( 303 ), using only a predetermined number (n) of these data considered effective ( 305 . 306 ). The combustion state controller for an internal combustion engine according to claim 1, further comprising: an air-fuel ratio feedback device ( 106 . 107 ) for controlling the closed-loop air-fuel ratio with respect to a target air-fuel ratio established in the lean air-fuel ratio range, the roughness calculating means (10) 303 to 307 ) the combustion roughness is calculated by statistical processing in a state that only the data in a predetermined established range of the torque variation is considered to be effective when the air-fuel ratio is controlled in the lean closed-loop region. A combustion state controller for an internal combustion engine according to claim 1 or 2, further comprising: an average value calculating means (14); 204 ) for calculating an average value of the torque change, wherein the roughness calculating means ( 303 to 307 ) the combustion roughness calculated based on the torque change of the effective predetermined sample inventory parameter in a state that only the data equal to or greater than the average value of the torque variation is considered to be effective. A combustion state controller for an internal combustion engine according to any one of claims 1 to 3, wherein said roughness calculating means (10) 303 to 307 ) a standard deviation or an average square of the torque change calculates a predetermined sample inventory parameter and the combustion roughness is determined based on the calculated standard deviation or the averaged square. Combustion condition controller for an internal combustion engine according to one of the claims 1 to 4, wherein a maximum value and a minimum value of the rotational speed to be obtained for every combustion in the internal combustion engine and the rotation change is calculated as the torque change based on the Difference between these values. A combustion state controller for an internal combustion engine according to claim 1, further comprising fol comprising: a measuring device ( 201 . 202 . 2030 . 2040 ) for establishing a predetermined range to have a center near a peak point of estimated rotational speed and measuring a rotational speed at the estimated peak point and the rotational speed at least at two points located before and after the estimated peak point in a predetermined range; and a peak value calculator ( 2050 . 2050A . 2060 . 2060a ) for correcting the rotational speed at the estimated peak point by comparing the amount of rotational speed between that at the estimated peak point and at the two points before and after the estimated peak point and calculating the peak value. A combustion state controller for an internal combustion engine according to claim 6, wherein said peak value calculating means (14) 2050 . 2050A . 2060 . 2060a ) calculates a difference between the rotational speed at the point before the estimated peak point and the rotational speed at the point after the estimated peak point, and the rotational speed corrects at the estimated peak point in accordance with the difference. A combustion state controller for an internal combustion engine according to claim 7, wherein said peak value calculating means (15) 2050 . 2050A . 2060 . 2060a ) makes the correction for the correction of the rotational speed larger at the estimated peak point when the difference between the rotational speed at the point before the estimated peak point and the rotational speed at the point after the estimated peak point is larger. A combustion state controller for an internal combustion engine according to claim 6, wherein said peak value calculating means (14) 2050 . 2050A . 2060 . 2060a ) calculates the difference between the rotational speed at the estimated peak point and the rotational speed at the point before the estimated peak point and the difference calculates between the rotational speed at the estimated peak point and the rotational speed at the point after the estimated peak point and then the rotational speed at the estimated peak point corrected in accordance with the ratio of these speed differences. A combustion state controller for an internal combustion engine according to claim 9, wherein said peak value calculating means (14) 2050 . 2050A . 2060 . 2060a ) makes the correction for the correction of the rotational speed at the estimated peak point larger when the difference between the two calculated rotational speed differences is larger. Combustion condition controller for an internal combustion engine according to one of the claims 6 to 10, wherein the crank angle range required for combustion the internal combustion engine, divided into six subregions, being adjacent two consecutive three subregions formed are allocated by dividing an area into six sub-areas are as a predetermined range for obtaining the peak value the maximum side of the speed and the peak value of the minimum side the speed respectively. A combustion state controller for an internal combustion engine according to any one of claims 6 to 11, wherein: 201 . 202 . 2030 . 2040 ) Measures a required time for each same crank angle interval in a predetermined range; and the peak value calculator ( 2050 . 2050A . 2060 . 2060a ) measures the measured value of the crank angle interval including the estimated peak value to thereby calculate a true peak value by comparing the magnitude of the measured values using the measured value of the crank angle interval including the estimated peak point and the measured values at the points before and after the estimated peak point , Combustion condition controller for an internal combustion engine according to one of the claims 6 to 12, wherein: a rotation change for every Combustion is calculated based on at least the maximum peak value the speed, and where the rotation change a given sample inventory parameter of statistical processing is suspended; and determines a combustion roughness is based on the statistical processing result. A combustion state controller for an internal combustion engine according to any one of claims 1 to 13, further comprising: a target roughener device ( 308 ) applied to an air-fuel ratio control system for feedback control of the air-fuel ratio with respect to the target air-fuel ratio established in the lean air-fuel ratio range, for setting the target value of the burns roughness on the basis of the operating condition of the internal combustion engine; and a desired air-fuel ratio learning device ( 310 to 313 ) for learning the lean target air-fuel ratio in accordance with the deviation between the combustion roughness and the target value during the air-fuel ratio feedback control. Combustion condition controller for an internal combustion engine according to Claim 14, wherein a learning value of the lean target air-fuel ratio, which is learned by the desired air-fuel ratio learning means, is saved in a memory. The combustion state controller for an internal combustion engine according to claim 1, further comprising: inventory parameter changing means (14); 3030 to 3110 ) for changing the sample inventory parameter in accordance with an operating state of the internal combustion engine when the statistical processing is performed by the roughness calculator. A combustion state controller for an internal combustion engine according to claim 16, wherein said torque change calculating means (16) 203 ) calculates a maximum value and a minimum value of a rotational speed for each combustion of the internal combustion engine, and calculates the rotational change serving as the torque variation based on the difference between the maximum value and the minimum value. The combustion state control apparatus for an internal combustion engine according to claim 16 or 17, wherein said inventory parameter changing means ( 3030 to 3110 ) makes the sample inventory parameter smaller when the deviation between the combustion roughness calculated by the roughness calculator and its set value is large. The combustion state control apparatus for an internal combustion engine according to claim 16 or 17, wherein said inventory parameter changing means ( 3030 to 3110 ) makes the sample inventory parameter smaller when the internal combustion engine is operated in a transient state. The combustion state control apparatus for an internal combustion engine according to claim 16 or 17, wherein said inventory parameter changing means ( 3030 to 3110 ) the sample inventory parameter changes in accordance with the operating range of the internal combustion engine. A combustion state controller for an internal combustion engine according to any one of claims 16 to 20, wherein said roughness calculating means (10) 307 ) performs statistical processing on reference value units using a certain sample inventory parameter as the reference value, successively accumulates the result of the statistical processing, and calculates the burn accuracy on the basis of an average value of the accumulated value. A combustion state controller for an internal combustion engine according to claim 21, wherein said roughness calculating means (10) 307 ) performs the statistical processing in reference value units, successively accumulates the result of the statistical processing, and calculates the combustion roughness based on the average value of the accumulated value when the sample inventory parameter is equal to or greater than a predetermined value. A combustion state controller for an internal combustion engine according to any one of claims 16 to 22, wherein said roughness calculating means (10) 307 ) calculates a standard deviation or an averaged square of the torque variation of a given sample inventory parameter and determines the combustion roughness based on the calculated standard deviation or the averaged square; and the inventory parameter change facility ( 3030 to 3110 ) the sample inventory parameter changes in accordance with the operating state of the internal combustion engine when the roughness calculator calculates the standard deviation or the averaged square. A combustion state controller for an internal combustion engine according to any one of claims 16 to 23, further comprising: a target roughener device (10); 308 ) applied to an air-fuel ratio control system for feedback control of the air-fuel ratio with respect to the target air-fuel ratio established in a lean range of the air-fuel ratio, for setting the target value of the Ver combustion roughness based on the operating state of the internal combustion engine; and a desired air-fuel ratio learning device ( 310 - 313 ) for learning the lean target air-fuel ratio in accordance with the deviation between the combustion roughness and the target value during the air-fuel ratio feedback control. Combustion condition controller for an internal combustion engine according to Claim 24, wherein the learning value of the lean target air-fuel ratio, which is learned by the desired air-fuel ratio learning means, is saved in a memory. A combustion state controller for an internal combustion engine according to claim 25, wherein said inventory parameter changing means ( 3030 to 3110 ) makes the sample parameter small when a vehicle battery is newly connected to the memory used for securing the lean target air-fuel ratio learning value, or when the lean target air-fuel ratio has not yet been learned. A combustion state controller for an internal combustion engine according to any one of claims 24 to 26, wherein the inventory parameter changing means (14) 3030 to 3110 ) makes the sample inventory parameter large as the number of learning processes of the lean target air-fuel ratio increases. Combustion condition controller for an internal combustion engine according to one of the claims 16 to 27, wherein: the combustion state controller with a smoother is provided for smoothing the calculated combustion roughness; and the degree of smoothing the smoothing device changed will be in accordance with the operating state of the internal combustion engine. A combustion state controller for an internal combustion engine according to claim 1, wherein: said roughness calculating means (14) 303 to 307b ) in calculating a combustion roughness, data in a predetermined set area and data outside the predetermined set area of the calculated torque change are weighted differently from each other. A combustion state controller for an internal combustion engine according to claim 29, further comprising: an air-fuel ratio feedback device ( 106 . 107 ) for controlling the closed-loop air-fuel ratio with respect to a target air-fuel ratio established in the lean air-fuel ratio range, the roughness calculating means (10) 303 to 307 ) the combustion roughness is calculated by statistical processing in a condition that the data in the predetermined established range of the torque variation and the data outside the predetermined established range of the torque variation are differently weighted when the air-fuel ratio is controlled in the lean region in the closed loop. The combustion state control apparatus for an internal combustion engine according to claim 29 or 30, further comprising: average value calculating means (14); 204 ) for calculating an average value of the torque change, wherein the torque change greater than the average value is used as the data in the predetermined established range and the torque change smaller than the average value is used as the data outside the predetermined established range , A combustion state controller for an internal combustion engine according to claim 31, wherein said roughness calculating means (10) 303 to 307b ) calculates a first roughness calculated from the data in the predetermined set-up area and a second roughness from the data outside the predetermined set-up area and the combustion roughness calculated by different weights of the first roughness and the second roughness. A combustion state controller for an internal combustion engine according to any one of claims 29 to 32, wherein said roughness calculating means (14) 303 to 307b ) uses a weighting constant that is variable with the engine operating condition. Combustion condition controller for an internal combustion engine according to Claim 33, wherein the weighting constant is set to a small value in a normal operating state of the engine of the internal combustion engine and on a big one Value in another engine condition including acceleration and a delay the internal combustion engine.