Method for operating an internal combustion engine

Abstract

A method for operating an internal combustion engine, in particular a gas engine having at least two cylinders, includes acquiring a cylinder-specific first cylinder signal (p.sub.max, E) from each cylinder. At least one combustion parameter (Q, Z) of the corresponding cylinder is controlled as a function of the first cylinder signal (p.sub.max, E), and a cylinder-specific reference cylinder value (p.sub.max, E) is set for the first cylinder signal (p.sub.max, E) for each cylinder. The at least one combustion parameter (Q, Z) of the cylinder is adjusted as a function of the deviation of the first cylinder signal (p.sub.max, E) from the reference cylinder value (p.sub.max, E), and the first cylinder signal (p.sub.max, E) tracks the reference cylinder value (p.sub.max, E).

Claims

1. A method of operating an internal combustion engine having at least two cylinders, said method comprising: acquiring a cylinder-specific first cylinder signal from each of the at least two cylinders via a sensor; and controlling a combustion parameter of each of the at least two cylinders as a function of the respective cylinder-specific first cylinder signal using a control device; wherein said controlling comprises setting a cylinder-specific reference cylinder value for the cylinder-specific first cylinder signal for each of the at least two cylinders via the control device such that at least one of NOx emissions and combustion efficiency of the at least two cylinders is identical or similar, and using the control device to adjust the combustion parameter of each of the at least two cylinders as a function of a deviation of the respective cylinder-specific first cylinder signal from a corresponding cylinder-specific reference cylinder value such that the respective cylinder-specific first cylinder signal tracks the corresponding cylinder-specific reference cylinder value.

2. The method according to claim 1, wherein the cylinder-specific first cylinder signal is at least one of: an internal cylinder pressure signal, a cylinder exhaust temperature signal, a nitrogen oxide emissions signal, and a combustion air ratio signal.

3. The method according to claim 2, wherein the cylinder-specific first cylinder signal is a maximum internal cylinder pressure of a combustion cycle signal.

4. The method according to claim 1, wherein the cylinder-specific reference cylinder value comprises a statistical variable of first cylinder signals of all of the at least two cylinders, and comprises a cylinder-specific offset from the statistical variable of the first cylinder signals.

5. The method according to claim 4, wherein the statistical variable of the first cylinder signals of all of said at least two cylinders is the arithmetic mean value.

6. The method according to claim 4, wherein the statistical variable of the first cylinder signals of all of said at least two cylinders is the median value.

7. The method according to claim 4, wherein the cylinder-specific offset is determined by a difference value characteristic mapping, the difference value characteristic mapping accounting for at least one of a power equivalent of an output power of the internal combustion engine and a charge air pressure of the internal combustion engine.

8. The method according to claim 7, wherein the difference value characteristic mapping further accounts for at least one of a charge air temperature and an engine speed of the internal combustion engine.

9. The method according to claim 4, wherein the cylinder-specific offset is determined as a function of at least one of: a cylinder pressure during a compression phase before ignition, an air mass equivalent, a center of combustion, a compression ratio, and an ignition delay.

10. The method according to claim 9, wherein the cylinder-specific offset is determined as a function of at least one deviation of a cylinder parameter from a mean value of the cylinder parameter of all of the at least two cylinders.

11. The method according to claim 10, wherein the cylinder-specific offset is determined from respective deviations of cylinder parameters using the following formula:
?m=a*?pverd+b*?air+c*MFB+d*?e+e*/?delay wherein ?pverd is a deviation of the cylinder pressure during the compression phase before ignition, flair is a deviation of the air mass equivalent, ?MFB is a deviation in the center of combustion, ?? is a deviation in the compression ratio and ?delay is a deviation in the ignition delay, and a, b, c, d, e are weighting coefficients for the deviations.

12. The method according to claim 1, wherein the combustion parameter is a fuel quantity for each of the at least two cylinders.

13. The method according to claim 12, wherein said adjusting of the combustion parameter comprises increasing the fuel quantity for the corresponding cylinder if the respective cylinder-specific first cylinder signal is smaller than the corresponding cylinder-specific reference cylinder value.

14. The method according to claim 12, wherein said adjusting of the combustion parameter comprises decreasing the fuel quantity for the corresponding cylinder if the respective cylinder-specific first cylinder signal is larger than the corresponding cylinder-specific reference cylinder value.

15. The method according to claim 12, wherein a fuel metering valve is provided for each of the at least two cylinders, wherein said adjusting of the combustion parameter comprises adjusting an open period for each corresponding fuel metering valve to adjust the fuel quantity for each respective one of the at least two cylinders.

16. The method according to claim 1, wherein the combustion parameter is an ignition point for each of the at least two cylinders.

17. The method according to claim 16, wherein said adjusting of the combustion parameter comprises setting the ignition point earlier for a corresponding one of the at least two cylinders if the respective cylinder-specific first cylinder signal is smaller than the corresponding cylinder-specific reference cylinder value.

18. The method according to claim 16, wherein said adjusting of the combustion parameter comprises setting the ignition point later for a corresponding one of the at least two cylinders if the respective cylinder-specific first cylinder signal is larger than the corresponding cylinder-specific reference cylinder value.

19. The method according to claim 16, wherein an ignition device is provided for each of the at least two cylinders, wherein the ignition point for the ignition device is set in degrees of crank angle before TDC.

20. The method according to claim 1, wherein said adjusting the combustion parameter comprises determining a specifiable overall engine target value.

21. The method according to claim 20, wherein said determining of the specifiable overall engine target value comprises determining the specifiable overall engine target value from a specifiable fuel-air ratio.

22. The method according to claim 21, wherein the specifiable fuel-air ratio is determined from at least one of a power equivalent of an output power of the internal combustion engine, a charge air pressure, and an engine speed of the internal combustion engine.

23. The method according to claim 21, wherein the specifiable fuel-air ratio is determined from at least one of an electrical power from a generator connected to the internal combustion engine, a charge air pressure, and an engine speed of the internal combustion engine.

24. The method according to claim 20, wherein the specifiable overall engine target value is determined as a function of at least one of: (i) a deviation of a power equivalent of an output power of the internal combustion engine from a specifiable target power equivalent, and (ii) a deviation of an engine speed of the internal combustion engine from a specifiable target speed of the internal combustion engine.

25. The method according to claim 1, wherein said controlling further comprises using the control device to monitor a combustion condition of each of the at least two cylinders, and determine whether the combustion condition is normal or abnormal with respect to a specifiable reference state, and said adjusting comprises only adjusting the combustion parameter of the corresponding one of the at least two cylinders if the combustion condition is determined to be normal.

26. The method according to claim 25, wherein the combustion condition is at least one of knocking, auto-ignition, and interruptions in combustion, and the combustion condition of each of the at least two cylinders is determined to be normal if no knocking, auto-ignition, or interruptions in the combustion are identified.

27. An internal combustion engine comprising: at least two cylinders; a sensor for acquiring a cylinder-specific first cylinder signal from each of said at least two cylinders; and a control device for controlling a combustion parameter of each of said at least two cylinders as a function of the respective cylinder-specific first cylinder signal; wherein said control device is configured to set a cylinder-specific reference cylinder value for the cylinder-specific first cylinder signal for each of said at least two cylinders such that at least one of NOx emissions and combustion efficiency of the at least two cylinders is identical or similar, and to adjust the combustion parameter of each of said at least two cylinders as a function of a deviation of the respective cylinder-specific first cylinder signal from a corresponding cylinder-specific reference cylinder value such that the respective cylinder-specific first cylinder signal tracks the corresponding cylinder-specific reference cylinder value.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) Further details and advantages of the present invention will now be provided with the aid of the accompanying description of the drawings, in which:

(2) FIG. 1 shows an exemplary representation of the dependency of the cylinder efficiency on NOx emissions from the cylinders of an internal combustion engine;

(3) FIG. 2 shows an exemplary representation of the tracking of cylinder-specific first cylinder signals on cylinder-specific reference cylinder values;

(4) FIG. 3 is a schematic diagram of an internal combustion engine with a plurality of cylinders and a control device for operating the internal combustion engine in accordance with an embodiment of the proposed method;

(5) FIG. 4 is a schematic diagram of 3 cylinders of an internal combustion engine and a control device for operating the internal combustion engine in accordance with an embodiment of the proposed method;

(6) FIG. 5 is a schematic diagram similar to FIG. 4 showing an internal combustion engine with a fuel-led combustion process;

(7) FIG. 6 is a schematic diagram of a proposed control device;

(8) FIG. 7 is a schematic diagram similar to FIG. 4 showing a further embodiment of the proposed method; and

(9) FIG. 8 is a schematic diagram of a control device of a further embodiment of the proposed method.

DETAILED DESCRIPTION OF THE INVENTION

(10) FIG. 1 shows, by way of example, the cylinder efficiencies ?.sub.cyl of two cylinders 2 of an internal combustion engine 1 (see FIG. 3) as a function of their respective NOx emissions E.sub.cyl as well as desired target values to be obtained with the proposed method for the NOx emissions E.sub.cyl and for the cylinder efficiencies ?.sub.cyl of all cylinders 2.

(11) The profile of a cylinder efficiency ?.sub.cyl to be obtained exhibits therein a non-linear dependency on the respective NOx emission E.sub.cyl of the corresponding cylinder 2. The different NOx emissions E.sub.cyl shown and the associated respectively different cylinder efficiencies ?.sub.cyl of the cylinders can in particular be caused by cylinder-specific differences in cylinder parameterssuch as different air charges, deposits and wear, centers of combustion, or mechanical tolerances of the cylinders 2.

(12) By means of the proposed method, these different cylinder-specific cylinder parameters can be taken into account, since for each cylinder 2, a cylinder-specific reference cylinder value p.sub.max is set for a first cylinder signal p.sub.max and a combustion parameter Q for the cylinder 2 (for example the quantity of fuel supplied to a cylinder) is set as a function of the deviation of the first cylinder signal p.sub.max from the reference cylinder value p.sub.max. The first cylinder signal p.sub.max tracks the reference cylinder value p.sub.max (see FIG. 2). In particular, in this manner, the cylinder-specific reference cylinder values p.sub.max of the cylinder 2 are adjusted in a manner such that the cylinder-specific NOx emissions E.sub.cyl to be obtained or the cylinder efficiencies ?.sub.cyl to be obtained for all cylinders 2 lie within a specifiable range or are essentially identical. All in all, by taking the cylinder-specific differences in the cylinder parameters into consideration, a general efficiency can be achieved over all cylinders 2 which is increased compared with the situation when these are not taken into consideration.

(13) FIG. 2 shows, for example, the profiles of a respective cylinder-specific first cylinder signal p.sub.max over the time t of three cylinders 2 of an internal combustion engine 1 (see FIG. 3). The cylinder-specific first cylinder signals p.sub.max here are the respective maximum internal cylinder pressures p.sub.max of the corresponding cylinders 2, each acquired over a combustion cycle of the corresponding cylinders 2. Cylinder-specific differences in the cylinder parameters such as air charge or combustion properties result in different profiles for the first cylinder signals p.sub.max. The proposed method now proposes providing or setting a cylinder-specific reference cylinder value p.sub.max for each cylinder 2, and the respective actual first cylinder signal p.sub.max tracks the corresponding reference cylinder value p.sub.max. In this manner, for example, despite different cylinder properties or cylinder parameters, the respective NOx emissions E.sub.cyl of the cylinders 2 or the cylinder efficiencies ?.sub.cyl for the cylinders 2 exhibit the same or similar values, and all in all can produce an increased overall efficiency over all cylinders 2 than when the differing cylinder parameters of the individual cylinders 2 are not taken into consideration. As shown in the drawing, from a time t.sub.1, the individual first cylinder signals p.sub.max track the respective cylinder-specific reference cylinder values p.sub.max, from which time t.sub.1 control in accordance with the proposed method occurs.

(14) The respective reference cylinder values p.sub.max in the example shown are constituted by the arithmetic mean p.sub.mean of the maximum internal cylinder pressures p.sub.max of all cylinders 2 and a cylinder-specific offset ?m. The respective offsets ?m in this case take account of the cylinder-specific differences in the cylinder parameters (for example air mass equivalent, centre of combustion, compression ratio, ignition delay).

(15) FIG. 3 shows an internal combustion engine 1 with three cylinders 2. A cylinder pressure sensor 4 is associated with each cylinder 2 in order to acquire a cylinder-specific first cylinder signal. The cylinder-specific first cylinder signal may be the profile over time of the internal cylinder pressure or the maximum internal cylinder pressure p.sub.max over a combustion cycle. The cylinder-specific first cylinder signal may also be a temporally filtered signal of the maximum internal cylinder pressure p.sub.max over a plurality of combustion cycles, for example over 10 to 1000 combustion cycles, preferably over 40 to 100 combustion cycles. The cylinder-specific first cylinder signal p.sub.max acquired from a cylinder 2 is transmitted via a signal line 14 to a control device 7. The control device 7 can also carry out the determination of the maximum internal cylinder pressure p.sub.max over a combustion cycle or temporal filtering of the maximum internal cylinder pressure p.sub.max over a plurality of combustion cycles.

(16) As will be described below, the control device 7 according to the proposed method, determines a respective cylinder-specific fuel quantity Q to be metered as a combustion parameter for the cylinder 2 which is transmitted to the corresponding fuel metering valve 3 via control lines 15. The fuel metering valves 3 dose the corresponding cylinder-specific fuel quantities Q into the cylinders 2 and thus the cylinder-specific first cylinder signals p.sub.max track the cylinder-specific reference cylinder values p.sub.max generated by the control device 7 according to the proposed method.

(17) FIG. 4 shows a diagrammatic block diagram of three cylinders 2 of an internal combustion engine 1 with an air-led combustion process. A fuel metering valve 3 is associated with each cylinder 2, wherein the fuel quantity Q supplied to the corresponding cylinder 2 can be adjusted by the respective fuel metering valve 3. A control device 7 controls the fuel metering valves 3, whereby the control device 7 outputs a respective cylinder-specific open period for the fuel metering valve 3 in the form of a cylinder-specific parameter t.sub.cyl.

(18) The fuel metering valves 3 in this example are port injection valves which have only a completely open and a completely closed position. When the fuel metering valve 3 is in the completely open position, a fuel in the form of a propellant gas is injected into the inlet tract of the cylinder 2 associated with the fuel metering valve 3. The open period of the fuel metering valve 3 can thus be used to set the fuel quantity Q for the respective cylinder 2.

(19) A cylinder-specific first cylinder signal p.sub.max is acquired from each cylinder 2 and supplied to the control device 7. In this regard, a cylinder-specific first cylinder signal p.sub.max corresponds to the maximum internal cylinder pressure of the corresponding cylinder 2 during a combustion cycle. In the example shown, the cylinder-specific first cylinder signals p.sub.max are supplied to a differential value processor 8 of the control device 7. The differential value processor 8 determines a difference value ?t.sub.cyl for each cylinder 2, or for each fuel metering valve 3, which is respectively added to a specifiable target value t.sub.g. Thus, a cylinder-specific open period is generated for each fuel metering valve 3 as a parameter t.sub.cyl.

(20) The specifiable overall engine target value t.sub.g in the example shown is determined from a specifiable fuel-air ratio ?, wherein the specifiable fuel-air ratio X is determined by an emission controller 5a from a power equivalent P of the output power of the internal combustion engine 1 (for example the electrical power measured for a generator connected to the internal combustion engine 1) and/or from a charge air pressure p.sub.A and/or from an engine speed n of the internal combustion engine 1. In addition to the fuel-air ratio ?, in a target value processor 6, the pressure p.sub.A and the temperature T.sub.A of the charge air, the pressure p.sub.G and the temperature T.sub.G of the fuel supply as well as the engine speed n of the internal combustion engine 1 may also be input. Furthermore, yet another flow parameter of the fuel metering valve 3 (for example the effective diameter of flow in accordance with the polytropic outflow equation or a Kv value) as well as fuel or combustion gas characteristics (for example the gas density, the polytropic exponent or the calorific value) can be input into the target value processor 6. The target value processor 6 then determines the specifiable target value t.sub.g, which corresponds to an overall engine basic open period value for the open periods of all of the fuel metering valves 3.

(21) By means of the difference value processor 8, a cylinder-specific open period offset or difference value ?t.sub.cyl is determined for each individual fuel metering valve 3. These cylinder-specific difference values ?t.sub.cyl are dependent on the deviation of the peak cylinder pressure p.sub.max of the respective cylinder 2 from the respective cylinder-specific reference cylinder value p.sub.max. The respective sum of the overall engine basic open period value t.sub.g and the cylinder-specific open period offset ?t.sub.cyl generates the target open period t.sub.ry, for the respective fuel metering valve 3 controlled by the drive electronics.

(22) Alternatively or in addition to using the maximum internal cylinder pressure p.sub.max as the cylinder-specific first cylinder signal, the use of the respective cylinder-specific cylinder exhaust temperature T.sub.E is indicated in dashed lines.

(23) In this manner, again, deviations in the cylinder-specific cylinder exhaust temperatures T.sub.E from the respective cylinder-specific reference cylinder values for the cylinder exhaust temperatures can be used to calculate the corresponding cylinder-specific open period offsets ?t.sub.cyl. The cylinder-specific cylinder exhaust temperatures T.sub.E may be used as an alternative, for example, when no internal cylinder pressure sensors 4 have been installed or also as a fall-back position if the cylinder pressure signals fail, in order to increase the availability of the internal combustion engine 1 in the case of a cylinder pressure sensor failure.

(24) FIG. 5 shows a block diagram similar to FIG. 4. In this case, the internal combustion engine 1 is powered by a gas-led combustion process. The specifiable overall engine target value t.sub.g in the example shown is determined by a controller 5b which can comprise a power controller and/or a speed controller. For the power controller, in addition to a power equivalent P for the output power of the internal combustion engine 1 (actual power), a specifiable target power equivalent P.sub.s (reference power) of the internal combustion engine 1 can serve as the input variable. For the speed controller, in addition to a respective actual engine speed n (actual speed) of the internal combustion engine 1, a specifiable target speed n.sub.s (reference speed) of the internal combustion engine 1 can serve as the input variable. In the controller 5b, an overall engine default value for the fuel mass flow m is determined, from which subsequently, in a target value processor 6 the specifiable overall engine target value t.sub.gfor example for the overall engine open period of fuel metering valves or for the overall engine default value for the ignition point of ignition devicesis determined.

(25) FIG. 6 shows a block diagram similar to FIG. 4, in which the control device 7 as well as the difference value processor 8 are shown in more detail. This representation shows details of the control procedure for just one cylinder 2 of the internal combustion engine 1. Other cylinders 2 of the internal combustion engine 1 are shown here as dashed lines.

(26) An internal cylinder pressure sensor 4 is associated with each cylinder 2. An internal cylinder pressure sensor 4 can thus acquire the profile of the internal cylinder pressure p.sub.cyl over a combustion cycle. A maximum acquired value processor 9 can determine the maximum internal cylinder pressure p.sub.max or the peak pressure of the respective cylinder 2 in the preceding combustion cycle.

(27) The peak pressures of all cylinders 2 are supplied to a mean computation processor 10 as cylinder-specific first cylinder signals p.sub.max. In the example shown, this mean computation processor 10 generates the arithmetic mean p.sub.mean from the cylinder-specific first cylinder signals p.sub.max and outputs it. In addition, a cylinder-specific offset ?m is computed in an offset processor 18 and output. The sum of the arithmetic mean p.sub.mean of the cylinder-specific first cylinder signals p.sub.max from all cylinders 2 and the cylinder-specific offset ?m in the example shown generates the cylinder-specific reference cylinder value p.sub.max which is supplied to the reference value controller 11.

(28) In the example shown, the cylinder-specific offset ?m is computed in an offset computation processor 18 from the internal cylinder pressure in the corresponding cylinder 2 before ignition p.sub.cyl (after closing an inlet valve associated with the cylinder 2 during the compression stroke) and from the center of combustion of the cylinder 2. In this manner, the internal cylinder pressure before ignition p.sub.cyl is either determined directly from the temporal profile of the internal cylinder pressure signal p.sub.cyl via a corresponding pressure computation processor 19 or from a load-dependent pressure determination characteristic mapping 20. The pressure determination characteristic mapping 20 here can contain appropriate values for the internal cylinder pressure before ignition p.sub.cyl, which are dependent on the load and/or the charge air pressure p.sub.A and/or the charge air temperature T.sub.A and/or the engine speed n of the internal combustion engine 1. The selection of the source for the value of the internal cylinder pressure before ignition p.sub.cyl is made by a pressure source switch 22. The determination of the center of combustion of the respective cylinder 2 is carried out in a center of combustion computation processor 21 in known manner from the temporal profile of the internal cylinder pressure signal p.sub.cyl.

(29) In general, the cylinder-specific offset ?m can be determined as a function of at least one of the following cylinder-specific cylinder parameters: air mass equivalent, center of combustion, compression ratio, ignition delay. Thus, the determination of the cylinder-specific offset ?m can be based on deviations of at least one of the respective cylinder parameters from the mean of this cylinder parameter over all cylinders 2.

(30) In the reference value controller 11, the deviation of the first cylinder signal p.sub.max of a cylinder 2 from the corresponding reference cylinder value p.sub.max is determined. Subsequently, a difference value ?t.sub.cyl is determined for the fuel metering valve 3 associated with the cylinder 2.

(31) The respective difference value ?t.sub.cyl for a fuel metering valve 3 associated with the respective cylinder 2 is then added to an overall engine specifiable target value t.sub.g, whereupon an open period for the fuel metering valve 3 is generated as a parameter t.sub.cyl. The specifiable target value t.sub.g is thus determined, as described in FIG. 4, from an emission controller of the internal combustion engine 1. It can basically also be determined from a power controller and/or from a speed controller (as described in FIG. 5) of the internal combustion engine 1.

(32) In the example shown, the respective difference value ?t.sub.cyl comprises a cylinder-specific pilot value t.sub.g, which is determined by a pilot value computation 12 from the charge air pressure p.sub.A and/or the charge air temperature T.sub.A and/or the engine speed n of the internal combustion engine 1. This respective pilot value t.sub.g can, for example, be determined by measurements during placing the internal combustion engine into operation and set out in a characteristic mapping.

(33) In general, the reference value controller 11 can, for example, be a P-, PI- or PID controller. However, other controller concepts and controller types may be used, for example a LQ controller, a robust controller or a fuzzy controller.

(34) In order to avoid unwanted consequences for the overall engine control, and in particular the emission controller 5a, the respective difference values ?t.sub.cyl are in addition provided with an equalization value t.sub.o from an equalization value processor 13. This equalization value t.sub.o, which is the same for all difference values ?t.sub.cyl, corresponds to the arithmetic mean of the difference values ?t.sub.cyl of all cylinders 2 and can be positive or negative. Thus, it is possible to apply the proposed method to internal combustion engines 1 which until now have been operated without cylinder balancing or only with a general controller, without this additional control having an impact on the overall engine control.

(35) FIG. 7 shows a diagrammatic block schematic similar to FIG. 4. However, in the illustrated embodiment of the invention, the ignition points Z from ignition devices 23 on or in the cylinders 2 rather than the fuel quantities Q for the cylinders 2 are set. The overall specifiable target value t.sub.g (overall default value) for the ignition point Z in this case is determined from an ignition point characteristic mapping 16, in which ignition point characteristic mapping 16 suitable values are presented for the overall default value t.sub.g as a function of the power or the power equivalent P and/or the charge air pressure p.sub.A and/or the charge air temperature T.sub.A and/or the engine speed n of the internal combustion engine 1. The respective parameter t.sub.cyl determined by the control device 7expressed in degrees of crank angle before TDCis sent to an ignition controller 17. The ignition controller 17 activates the respective ignition device 23 at the given ignition point Z. In this manner, in this example, the ignition point Z of a cylinder 2 is set earlier with respect to the overall default value t.sub.g if the peak cylinder pressure p.sub.max of the cylinder 2 (first cylinder signal) is smaller than the reference cylinder value p.sub.max and the ignition point Z of a cylinder 2 is set later with respect to the overall default value t.sub.g if the peak cylinder pressure p.sub.max of the cylinder 2 is larger than the reference cylinder value p.sub.max.

(36) FIG. 8 shows a diagrammatic block schematic of a further embodiment of the invention which is similar to that of FIG. 6, but the ignition points Z of the ignition devices 23 on or in the cylinders 2 rather than the fuel quantities Q for the cylinder 2 are set. In this example, the nitrogen oxide emissions E.sub.cyl of a cylinder 2 are respectively acquired over a combustion cycle from a NOx probe 24 and sent to an analytical unit 25. From the temporal profile of the nitrogen oxide emissions E.sub.cyl over a combustion cycle, the analytical unit 25 determines a filtered emission value which is sent as the cylinder-specific signal E to the reference value processor 10. The reference value processor 10 generates the median E.sub.median from the cylinder-specific signals E from all cylinders 2 and outputs it. In addition, in an offset processor 18, a cylinder-specific offset ?m is computed and output. The sum of the median E.sub.median and the cylinder-specific offset ?m in the example shown generates the cylinder-specific reference cylinder value E, which is sent to a reference value controller 11.

(37) The cylinder-specific offset ?m in the example shown is determined in an offset processor 18 by reading out a difference value characteristic mapping 26, in which appropriate values for the offset ?m for the corresponding cylinder 2 are recorded as a function of the power P and/or the charge air pressure p.sub.A and/or the charge air temperature T.sub.A and/or the engine speed n of the internal combustion engine 1. Here, the values deposited in the difference value characteristic mapping 26 for the cylinder-specific offsets ?m of the cylinders 2 were determined on a test rig.

(38) In the reference value controller 11, the deviation of the cylinder-specific signal E from the reference cylinder value E is determined and as a function thereof, a difference value ?t.sub.cyl is determined for the ignition point Z of an ignition device 23 associated with the corresponding cylinder 2. The respective difference value ?t.sub.cyl is then added to the overall engine specifiable target value t.sub.g, whereupon an ignition point Z is generated in degrees of crank angle before TDC as the parameter t.sub.cyl and sent to an ignition controller 17. The ignition controller 17 activates the ignition device 23 (for example a spark plug) at the given ignition point Z. The specifiable target value t.sub.g in this regard is determined from an ignition point characteristic mapping 16 as described in FIG. 7.