WO1991011599A1

WO1991011599A1 - Process for detecting misfiring in an internal combustion engine

Info

Publication number: WO1991011599A1
Application number: PCT/DE1991/000013
Authority: WO
Inventors: Martin Klenk; Winfried Moser
Original assignee: Robert Bosch Gmbh
Priority date: 1990-01-26
Filing date: 1991-01-09
Publication date: 1991-08-08
Also published as: DE4002209A1; DE4002209C2

Abstract

A process for identifying a misfiring cylinder in an internal combustion engine with Z cylinders comprises the following steps. For each cylinder z (z of 1 - Z), a time-interval T(z) in which the crankshaft rotates through a given crank angle in the combustion cycle of the cylinder is measured, and for each cylinder the time difference DT(z) = T(z) - T(z-1) is calculated. The existence of misfiring is detected by any cylinder-nonspecific process. The first time that misfiring occurs during the process, the mean differential value for each cylinder is calculated and compared with a threshold value. When a predetermined mean value-threshold value criterion is met for one cylinder, this cylinder is considered to be misfiring. A misfiring cylinder can be reliably identified by this process without using special sensors.

Description

Misfire detection in an internal combustion engine

Misfires occasionally occur in internal combustion engines. The invention relates to a method for determining in which cylinder the misfires occur. Several cylinders can also be affected, which is then determined accordingly.

If combustion misfires occur, this leads to the unburned fuel / air mixture entering the catalytic converter with which the exhaust system of internal combustion engines is usually equipped nowadays. The mixture burns in the catalytic converter, which can lead to the destruction of the catalytic converter and even to a vehicle fire due to the temperature increase achieved thereby. Efforts are therefore made to identify a cylinder with combustion misfires in order to be able to disconnect this cylinder from the fuel supply.

If the term "misfires" is shortened below, this means combustion misfires. These can be due to various causes, e.g. B. by misfiring or by excessive emaciation or over-greasing of the mixture. State of the art

Extremely cylinder-selective misfire detection methods are known, as are extremely unselective methods with any intermediate stages. If the combustion pressure is measured in each cylinder, a decision can be made with a high degree of confidence in which cylinder misfire occurs. This advantage is offset by the disadvantage that separate pressure sensors are required. In contrast, there are e.g. B. Methods that measure the temperature of the catalyst and examine whether the temperature is above an operating point dependent threshold. Such methods have the advantage that only a very cheap and only temperature sensor is required, which is also often available in series. The disadvantage is that it is completely unclear which cylinder has misfires.

Misfire detection methods are also known which compare the angular velocity values of the crankshaft with one another, which are measured within predetermined angular ranges. These methods have the advantage that no separate sensor is required, but there is the problem that angular velocities of the crankshaft depend not only on combustion processes, but also in particular on forces which are transmitted from the roadway to the crankshaft via the wheels and the drive train . In order not to erroneously identify misfires in a cylinder due to external influences, it is known (DE-A-36 15 547) to recognize misfires in a cylinder only when the misfire detection conditions have been met several times in succession.

From the above it is clear that there is generally the problem of specifying a method for determining a cylinder with misfires in an internal combustion engine, which is designed in such a way that the simplest possible means, but as reliably as possible, of which of Z cylinders misfires can be identified having. Presentation of the invention

A first method according to the invention for determining a cylinder with misfires in an internal combustion engine with Z cylinders is characterized in that

- For each cylinder z (z from 1 - Z) a time period T (z) is measured, within which the crankshaft rotates by a predetermined crank angle in the combustion cycle of the cylinder, and for each cylinder the time difference value DT (z) =

T (z) - T (z - 1) is formed,

- it is determined by any cylinderselective method whether misfires occur,

and then, if misfires occur during the course of the method, a differential averaging with threshold comparison is carried out for each cylinder, and then, if a predetermined average threshold criterion for a cylinder is met, a decision is made that this cylinder has misfires. This method has a very selective effect, in particular because it uses a mean value difference. This difference value averaging contributes to the fact that individual outliers in the time measurements do not cause false signals. Despite the averaging, clear distinctive signals are obtained due to the formation of the difference. However, this only applies to periods in which the difference values change, i.e. only if misfires occur suddenly and continuously after a misfire-free engine run. Are misfires from the start of the procedure, e.g. B. from the start of an engine, the method is not applicable, since in this case the difference values remain essentially zero permanently. The process is therefore only carried out completely if it is determined in the course of the process that misfires occur for the first time. If it is found in the test just mentioned that misfires have existed since the start of the method, a conventional angular acceleration method can be used to determine a cylinder with misfires. However, in the case just mentioned, it is preferred to proceed in such a way that

- if it is found that misfires since the beginning of the

Method are present, for determining the cylinder or cylinders with misfires that

- If the time difference value DT (a) or the averaged time difference value MW_DT (a) (a within 1 - Z) exceeds a predetermined threshold value SW, it is decided that this cylinder has a misfire.

This method uses signals that are also used for the method that provides particularly precise results when misfires only occur during the course of the method and can therefore be used in a particularly advantageous manner together with it. However, it can also be used alone or together with another method.

With the methods mentioned in the previous paragraph, be it a known method or the preferred method, there is the problem that the decision certainty is not too high. As already mentioned, a final decision is therefore only made in the prior art when the misfire detection condition for a cylinder has been fulfilled more than once. However, it is more advantageous to carry out the method according to claim 3, namely

when it is decided which misfire cylinder has to disconnect this cylinder from the fuel supply,

- subsequently to check whether the misfire detection method delivers a substantially unchanged misfire detection signal, and

then, if the misfire detection signal is essentially unchanged, to decide that the cylinder which actually has misfires has been correctly disconnected from the fuel supply, - On the other hand, if the misfire detection signal changes significantly, add fuel to the cylinder concerned and check again which cylinder triggered the misfire detection.

This method can also be used together with the method of claim 1, but in connection with this method it will very rarely lead to a revision of the decision made.

drawing

1 shows a flow diagram for explaining a method for determining a cylinder with misfires in an internal combustion engine;

FIG. 2 diagram relating to the time course of the smallest mean value of time difference values for a cylinder a with misfires;

3 shows a diagram which is time-correlated with the diagram of FIG. 2 and represents the time profile of a mean difference value and an overall mean value; and

Fig. 4 flow chart for explaining a sub-method with which a decision made regarding a cylinder is checked with misfires. Description of exemplary embodiments

The sequence according to FIG. 1 takes place repeatedly within a main program during the operating period of an internal combustion engine. In step s1, a period of time T (z) is measured for every cylinder z (z from 1-Z) every 720 ° crank angle, within which the crankshaft rotates by a predetermined crank angle in the combustion cycle of the cylinder. Is measured e.g. B. between TDC of the piston in the combustion cycle and 720º / Z after TDC, that is over a range of 180 ° from TDC in a four-cylinder engine or from 120 ° from TDC in a six-cylinder engine. The measurement takes place every 720º / Z crank angle for one cylinder each. Immediately after each measurement for a cylinder, the time difference value DT (z) becomes for this cylinder. = T (z) - T (z - 1). If this value is less than zero, this means that the crankshaft runs faster in the combustion cycle of the current cylinder than in the combustion cycle of the cylinder in which combustion took place previously. It is then very unlikely that there is a misfire in the current cylinder. Every negative time difference value can therefore also be set to zero. In step s2, a misfire detection signal is retrieved, as is provided by any cylinder-selective misfire detection method. In step s3 it is checked whether misfires are present. If the engine runs without interruption, processing of other routines in the main program continues.

On the other hand, if step s3 shows that misfires exist, the moving average MW_DT (z) of the time difference values DT (z) is first formed in a step s4 according to the following equation:

MW_DT (z) = k x MW_DT (z) _alt + (1 - k) DT (z)

Here, k is a filter factor <1, which is preferably above 0.9, in particular between 0.93 and 0.95. It is pointed out that the moving average can also be formed differently, e.g. B. by deleting the oldest value in a series of values and adding the new value.

It is then checked (step s5) whether the misfires are present from the start of the method. This can check z. B. take place in that a start detection flag is set to zero with the start of the internal combustion engine. When the method illustrated by FIG. 1 returns, this flag is set to one. If step s5 is already reached during the first run of the method after the engine is started because of misfires, the flag is still at zero. If this is the case, it is decided in a step s6 that the cylinder a has misfires for which the averaged time difference value MW-DT (a) is above a threshold value SW. In a step s7, the cylinder a is disconnected from the fuel supply.

After the fuel supply has been disconnected, a decision can be checked between two marks A and B, which is explained below with reference to FIG. 4. Then the next routine in the main program is continued.

If it is found in step s5 that misfires first occur during the course of the method, step s8 follows, which is carried out every 720 ° / Z crank angle for one of the Z cylinders. The change in the moving average formed in step s4 is examined. The moving average will fluctuate only slightly as long as there are no misfires in the cylinder in question. However, if misfires occur, the mean initially runs quickly and then slower and slower to a new value. The change, which, as just mentioned, initially has a large value, is calculated as follows:

DMW_DT (z) = MW_DT (z) - MW_DT (z) _old If a mean value difference value DMW_DT (z) <0, it is set to zero.

After completing these steps, the old mean gets the value of the new mean.

In a step s9, which is carried out only every 720 ° crank angle, the mean value of the mean difference values is formed, which is done by summing the latter values and dividing the sum by the number Z of cylinders. Accordingly, the following applies to the overall mean: MW DMW = DMW_DT (Z)) / Z

This overall mean value forms the basis for deciding which cylinder a has misfires. It is the cylinder for which the following applies:

DMW_DT (a)> MW_DMW

If this condition is met for several cylinders, this is a sign that misfires occur in all of these cylinders. For all these cylinders, the steps already mentioned follow from step s7.

The time course of various of the signals just mentioned can be seen in a time-correlated manner from FIGS. 2 and 3. No dropouts should occur up to a point in time ZP_A. The mean value difference value MW_DT for an individual cylinder a, in which misfires then occur from the time ZP_A, is then very low. The mean value increases from this point in time, but this is only examined from a point in time ZP_E at which the presence of misfires is detected with the aid of another method. The detection time ZP_E must not be so far behind the occurrence time ZP_A that the mean value MW_DT (A) is already largely at its new value is swinging. Then no significant change over time in the mean value just mentioned could be determined.

3 shows the change over time of the mean value MW_DT (a), that is to say the change over time of the value DMW_DT (a). The entire mean value MW_DMW is also shown. 2, the mean value difference value DMW_DT (a) is clearly above the total mean value MW_DMW. This is a sign that misfires are present in cylinder a.

The threshold value for the comparison with the mean value difference values can be used instead of the explained overall mean value MW_DMW, for example a predefined threshold value. However, the procedure according to the exemplary embodiment has the advantage that the threshold value is continuously adapted to operating conditions and that, after the occurrence of misfires, it changes in time in a similar way to how a mean value difference value to be compared changes.

The decision verification method between the marks A and B already mentioned in the description of FIG. 1 will now be explained with reference to FIG. 4. In step s11, the misfire detection signal is called up by any cylinder-selective misfire detection method, which was already mentioned above with reference to step s2. If the measure in step s7 regarding the disconnection of cylinder a from the fuel supply was correct, the misfire detection signal must not have changed. However, if the wrong cylinder was no longer supplied with fuel, this cylinder now also shows misfires, in addition to the cylinder that triggered the determination process. A significant change in the drive detection signal, which may be determined in step s12, is therefore a sign that the wrong cylinder has been switched off. If this is the case, the fuel supply is in a step s13 released for cylinder a again. Routines of the main program follow the following label B, within which the described determination method is also called up again.

Wrong decisions can also be avoided within certain limits in that a cylinder a is only separated from the fuel supply when it has been found in several runs of the determination method according to FIG. 1 up to step s10 that this cylinder has misfires. However, this method does not work as quickly as that described with reference to FIG. 4, and there is also the problem that a predetermined number of misfire detections for a cylinder can also be achieved accidentally within a predetermined period of time, namely when e.g. B. Faults from the roadway are transmitted randomly in such strength and time sequence that these faults are quickly assigned to a single cylinder in succession. Wrong decisions based on such a situation are practically excluded by the method according to FIG. 4.

After a cylinder is disconnected from the fuel supply, a healing attempt can be made from time to time in a known manner, i. H. the cylinder is supplied with fuel again and it is examined whether misfires still occur.

The measure of stopping the supply of fuel to a misfiring cylinder is not the only measure that can be taken in such a case. For example, the cylinder can also be disconnected from the air supply and / or a lambda control can be switched to control. In the method explained with reference to FIG. 1, it was assumed that first an average of time difference values is formed, an average value difference value is calculated from these average values, and then this average value difference value is compared with a predetermined threshold value. However, it can also be done in such a way that differences are formed from the time difference values, these differences are compared with a threshold value and only then is averaging carried out, which in this case takes place by counting how often within a predetermined number of combustion cycles for a respective cylinder a difference exceeds a predetermined threshold value, the number determined in this way exceeds a threshold number, this is the sign of misfires in the relevant cylinder. Ultimately, the result is the same sequence as in the method explained with reference to FIG. 1, but with a different order of the calculation steps and therefore also different details of the calculation steps.

Claims

1. Method for determining a cylinder with misfires in an internal combustion engine with Z cylinders,

characterized in that

T (z) - T (z - 1) is formed,

- it is determined by any cylinderselective method whether misfires occur,

- and if misfires occur during the course of the method, a differential averaging with threshold comparison is carried out for each cylinder, and then if a predetermined average threshold criterion for a cylinder is met, it is decided that the cylinder Cylinder misfires.

2. The method according to claim 1, characterized in that the difference value averaging with threshold value comparison and the checking of the criterion takes place in that

an average value MW_DT (z) of the time difference values DT (z) is formed for each cylinder z (z from 1-Z), a mean difference value DMW_DT (z) is continuously formed for two successive mean values for each cylinder,

- and then when a mean difference value DMW_DT (a) for

a cylinder a is above a predetermined threshold value, it is decided that this cylinder a has misfires.

3. The method according to any one of claims 1 or 2, characterized in that the threshold value for the comparison with the

Mean value difference values DMW_DT (z) is continuously formed by calculating the total mean value MW_DMW of the mean value difference values for all cylinders z (z from 1 - Z) every 720 ° crank angle.

4. The method according to claim 1, characterized in that the difference averaging with threshold value comparison and the checking of the criterion are carried out in that

the difference between two successive time difference values is continuously formed for each cylinder,

counting how often this difference between two successive time difference values for a cylinder is above a threshold value within a predetermined number of combustion cycles for this cylinder,

- and then if the number counted for a cylinder a

Threshold number exceeds, it is decided that this cylinder has a misfire.

5. A method for determining a cylinder with misfires in an internal combustion engine with Z cylinders, in particular method according to one of the preceding claims, characterized in that

- If it is determined that misfires have occurred since the start of the method, the cylinder or the cylinders with misfires is determined in such a way that - If the time difference value DT (a) or the mean time difference value MW_DT (a) (a within 1 - Z) exceeds a predetermined threshold value SW, it is decided that this cylinder has a misfire.

6. The method according to any one of the preceding claims, characterized in that

when it is decided which cylinder a has misfires, this cylinder is disconnected from the fuel supply,

it is then checked whether the method for cylinder-selective misfire detection delivers an essentially unchanged misfire detection signal, and

when the misfire detection signal is essentially unchanged, a decision is made that the cylinder which actually has misfires has been disconnected from the fuel supply,

- On the other hand, if the misfire detection signal changes significantly, the affected cylinder is supplied with fuel again and it is checked again which cylinder triggered the misfire detection.