Method and device for controlling a multicylinder internal combustion engine

Abstract

In a method for controlling a multicylinder internal combustion engine, where a cylinder-equalization control is performed to adjust cylinder-specific torque contributions to the overall torque of the internal combustion engine, it is provided for a functioning monitoring to be carried out for the cylinder-equalization control, and, in response to disturbances in the area of the cylinder-equalization control, for at least one error signal to be generated, indicating the existence of a disturbance.

Description

FIELD OF THE INVENTION

[0001] The present invention is directed to a method and a device for controlling a multicylinder internal combustion engine in accordance with the definition of the species set forth in the independent claims.

BACKGROUND INFORMATION

[0002] In control devices and, respectively, methods of the species for controlling multicylinder internal combustion engines, a cylinder equalization control is provided for adjusting and/or equalizing cylinder-specific torque contributions to the overall torque of the internal combustion engine. The internal combustion can be a gasoline or diesel engine, for example. There is a special need for a cylinder-equalization function in direct-injection gasoline engines. In this respect, increased irregular running can occur due to a low compatibility to tolerances of the combustion process in stratified operation and/or due to the tolerances of the high-pressure injectors used or in the distribution of the fuel to the individual cylinders. Age-related changes in the flow characteristic of the high-pressure injectors can also play a role. A cylinder equalization control can effect an improvement in running smoothness.

[0003] Within the framework of cylinder equalization, the active torque contribution (actual torque) of a particular cylinder is detected in relation to its preceding ignition, i.e., not as an absolute torque determination, based on an analysis of the time characteristic of the crankshaft or camshaft rotation. A torque correction can be effected, with the goal of equalizing cylinder-specific torque contributions, by intervening in at least one combustion-determining parameter, in particular in the injected fuel quantity, the ignition point timing in the case of a gasoline engine, in the exhaust-gas recirculation rate, or in the injection position. The term “injection position” refers here to the angular position of an injection pulse with respect to a reference point, for example to the top dead center of the piston of a cylinder in its power stroke.

[0004] In practice, those cylinder-equalization controls, in which so-called segment periods are recorded to analyze the time characteristic of the moment of rotation of the crankshaft or camshaft, have proven to be particularly worthwhile. Segment periods are the times it takes for the crankshaft or camshaft to cover a predefined angular range assigned to a particular cylinder. The more smoothly the engine is running, the more negligible the differences are between the segment periods of the individual cylinders. A measure of the engine's irregular operation is able to be derived from the mentioned segment periods. The known methods provide for an individual control to be assigned to each cylinder of the internal combustion engine and for the individual control to receive a cylinder-specific, irregular-operation actual value, as an input signal. On the basis of the irregular-operation actual values, a control setpoint value is determined. For example, the irregular-operation values of a plurality of cylinders can be averaged to produce a control setpoint value. On the output side, the controls influence the cylinder-specific injection times and, thus, the cylinder-specific torque contributions such that the cylinder-specific actual values approach the setpoint value. Proven cylinder-equalization controls are described, for example, in EP 140 065 or in DE 198 28 279. With regard to the generation of irregular-operation values, the disclosure of the last-mentioned laid-open print is expressly incorporated in this Application.

[0005] The object of the present invention is to provide improvements in the area of cylinder-equalization control.

SUMMARY OF THE INVENTION

[0006] In accordance with the present invention, improvements are achieved in the area of cylinder-equalization control in that the functioning of the cylinder-equalization control is monitored, and in that, in response to disturbances in the area of the cylinder-equalization control, at least one error signal is generated, indicating the existence of a disturbance. This makes it possible to recognize whether a cylinder-equalization control is actually performing its designated function in the manner intended.

[0007] When it comes to direct-injection gasoline engines, for example, conventional cylinder-equalization controls are not able to ensure that a controller intervention measure, such as enriching the combustion mixture via excessively proportioned fuel metering, in fact actually achieves the desired effect, i.e., leads to the desired increase in the torque contribution. Whether the desired effect is achieved may be recognized by functioning monitoring in accordance with the present invention, and appropriate remedial measures may be introduced in response to the at least one error signal. The functioning monitoring may be quasi wrapped around the cylinder-equalization control to monitor it for proper functioning. In this manner, controller stability and controller action may be ensured in the context of the cylinder equalization.

[0008] In the context of the cylinder-equalization control, one advantageous embodiment provides for at least one combustion-influencing performance quantity to be controlled on an individual basis, within a defined operating range, for each cylinder of the internal combustion engine. This is preferably the fuel metering, i.e., the quantity of the fuel to be injected, which, particularly in the stratified operation of gasoline engines, virtually exclusively determines the cylinder-specific torque. It is provided for an error signal to be generated when at least one cylinder-specific control reaches a boundary or a limit of the defined operating range. Thus, when the control goes beyond the operating range or the operating width, this is used as an indication of a possibly existing problem. It is assumed here that the normal task of a cylinder-equalization control is to correct the unavoidable torque differences that occur between the individual cylinders in normal operation of the internal combustion engine. As mentioned, these torque differences may be caused, for example, by differences in the flow characteristics of the individual high-pressure injectors that are due to tolerances. At present, it is not unusual to have typical tolerances in the range of 20 to 25% around an average injection value. Of equivalent magnitude are the torque errors to be expected because of scattering high-pressure injectors. In the context of the original application of an internal combustion engine, this is taken into consideration by properly rating the operating width range of the control. In this context, for example, an intervention stop, which is applied to twice the level as the expected maximum torque error, may be set for each controller. This provides the control with a certain safety reserve or ruggedness, and a normal controller operation is ensured. Thus, the operating range of the control is dimensioned to be ample enough to ensure that customary tolerances lie within the operating range and that a certain safety distance from the limits of the operating range remains. If, at this point, in at least one control, an operating range limit is reached (i.e., a controller runs to its preset intervention stop), then, obviously, the goal of cylinder-equalization control, namely the elimination of angular-acceleration contributions of the individual cylinders, is no longer attainable using the available, limited intervention of the control. This leads to the generation of an error signal.

[0009] One embodiment provides that, in response to an error signal, a defined change in the control characteristic of the control initially follows. In particular, the closed-loop gain is canceled in a suitably predefined manner. Redefining the controller performance in accordance with the embodiment, which may preferably be undertaken automatically and during operation of the internal combustion engine by changing the controller parameterization, advantageously considers the realization that the causes of a controller stop may lie in different areas. On the one hand, errors can occur in the area of the control itself (denoted here as controller defect), and, on the other hand, in the area of the function to be controlled (denoted here as cylinder defect). These types of errors can occur alternatively or cumulatively. As a rule, a controller defect can be viewed as “the lesser evil”.

[0010] For example, an instability of the control due to unpredictable parameter changes within the control system can occur, which is the cause of the control stop being reached. Thus, for example, changes may have occurred in the frictional torque of the engine. Should, in the area of the control, essentially only one, for example, aging-related system parameter change be the cause of the effected controller stop, then this may be considered by redefining the controller performance, in particular of the controller loop gain, thereby enhancing the ruggedness of the control. As a rule, once this adjustment is made, fluctuations will again lie within the control's operating range, including the redefined controller loop gain, and the control will again function without faults.

[0011] One preferred embodiment provides that the described adaptation of the controller characteristic, in particular by canceling the controller loop gain, be carried out at least one additional time when, following a first-time change, a controller stop again occurs, and an error signal is thereby generated. Thus, subsequent to a first change in the controller characteristic, as a response to a new error signal, the controller characteristic is changed at least one additional time, in particular through a subsequent cancellation of the controller gain. This can be set back, for example, in a first step to half, in a next step then to a third, subsequently to a fourth, etc., of the controller loop gain originally set. By successively reducing the controller loop gain to a predefinable final value, the control, which usually works with individual PI controllers, becomes more rugged, but also dynamically slower, which is taken into consideration.

[0012] Actual cylinder defects also come under consideration as the cause of a controller stop. Along the lines of the present application, these are understood, in particular, as mechanical defects associated with the injection and/or the ignition of a cylinder. It can happen, for example, that the torque expected from a cylinder is no longer able to be achieved, even via a (limited) enriching, i.e., a heavier fuel metering, because, for example, due to more pronounced changes in the valve flow rate and/or in the valve jet, the mixture is no longer able to be sufficiently introduced into the cylinder or properly ignited. For cases of this kind, in particular, it may be provided that, as a response to an error signal, preferably following an at least one-time change in the controller characteristic, the control is switched off by the functioning monitoring. In this context, the control characteristics are expediently reset to the originally defined operating range. Thus, the control is deactivated by the efficiency monitoring and restored to its original state.

[0013] Once the underlying defect is remedied, the control function is then able to immediately intervene with its original effectiveness.

[0014] To permit appropriate remedies promptly in response to the occurrence of defects, it is preferably provided for an external warning signal to be output as a response to an error signal, preferably following an at least one-time change in the controller characteristic. This may activate, for example, a preferably optical warning display that the operator is able to directly perceive and/or generate an entry to this effect in a memory of a diagnostic system, so that, in the context of a subsequent diagnosis, the defect that has occurred and, in some instances, been precisely defined, may be eliminated.

[0015] It is also possible that, as a response to an error signal, preferably following an at least one-time change in the control characteristic, the control is stopped by the functioning monitoring, the active controller characteristics being retained. Here, as well, preferably by outputting an appropriate external warning signal, the existence of a defect is indicated, and, thus, a request is made to remedy the defect.

[0016] Thus, the present invention renders possible, in particular, a multi-step, i.e., staggered reaction to problems that arise in the cylinder-equalization control. In a first step, in some instances to be repeated several times, the control characteristics are modified by suitable new controller dating that is able to be performed automatically and during operation of the internal combustion engine, such that ageing-related changes in the control system are considered, above all. Not until after this adaptation of the controller characteristics is made, when errors are still being indicated by the functioning monitoring, is it assumed that there are actual defects in the cylinder area possibly requiring a repair intervention. The first step may be executed, unnoticed by the operator, while the second step expediently generates warning signals, indicating a necessary maintenance or repair.

BRIEF DESCRIPTION OF THE DRAWING

[0017] These and other features are delineated in the claims, and in the Specification and drawing, as well. The only figure of the drawing shows an experimentally determined measuring diagram for elucidating the performance of the functioning monitoring of a cylinder-equalization control.

[0018] In FIG. 1, three diagrams are plotted over a common time axis (X-axis) to exemplify the time characteristic of a preferred specific embodiment of a functioning monitoring for a cylinder-equalization control, in the context of gasoline direct-injection engine equipped with four cylinders. The assigned values of the Y-axes were determined experimentally, the operation of the internal combustion engine being at approximately 1,400 rpm in stratified operation (15%).

[0019] The four curves K1, K2, K3, K4 in the top diagram indicate cylinder-specific angular-acceleration values K of the crankshaft of the internal combustion engine as Y-values. The setpoint value (control setpoint of the cylinder-equalization control) corresponding to a highest running smoothness lies, in the context of the selected definition, at value K1=K2=K3=K4=0. These ideal values are approximately set in the first five seconds of the plotting time period by the cylinder-equalization control. A value deviating from zero for at least one of the angular-acceleration values K1 through K4 signifies that a control intervention in the cylinder-equalization control is necessary.

[0020] In the middle diagram, control intervention factors R for the four PI controllers of the cylinder-equalization control assigned to the four cylinders, are plotted on the Y-axis. The ordinate values R representing a controller intervention are to be understood as factors plotted against the injection time at the particular cylinder. Thus, for example, a value of R1=1,2 signifies that, at cylinder 1, the injection time is increased by the control by 20% over a reference injection time, which ideally corresponds to a 20% increase in the injected fuel quantity. Consequently, values of R>1 correspond to an enriching of the mixture beyond the average mixture composition, while values of R<1 signify a leaning of mixture. Preferably, the overall control is set such that the total injection quantity essentially remains the same per unit of time.

[0021] As mentioned at the outset, it is not unusual to have typical tolerances in the range of 20 to 25% around an average injection value. This variance may be caused, in particular, by differences in the flow characteristics of the individual high-pressure injectors that are due to tolerances. To ensure that the functioning monitoring does not already interpret fluctuations in this range as an indication of the existence of a defect going beyond this in the cylinder area and/or in the control, in the illustrated example, the operating range of the controls is adjusted by appropriate controller dating, such that the controls reach their controller stop at 30% deviation from intervention factor 1® value=1.3). Between the expected maximum tolerance of, for example, 25% and this intervention stop EA, a certain safety reserve or ruggedness still exists, which ensures a normal controller operation in spite of possibly existing tolerance-related deviations among the individual injections.

[0022] The three rectangular-curve type lines in the bottom diagram represent control bits B of the digital engine control, which also includes the controllers. In this context, top curve B1 represents the controller timing, and the essentially regularly occurring rectangular pulses of about 200 millisecond length represent the corresponding controller clock-pulse bits, which prompt a clocked updating of the controllers.

[0023] Middle curve B2 represents the error indication of the efficiency monitoring. A corresponding error bit, as occurs, for example, at instant t=7.2 seconds, thus corresponds to an error signal of the efficiency monitoring and occurs under certain conditions which are already mentioned at the outset and are explained in further detail below.

[0024] Bottom line B3, coinciding predominately with the abscissa, represents a switch-off function for the control via the functioning monitoring and is defined such that, when a bit is not set, the control (line on abscissa) is switched on, and, when a bit is set, the control is switched off. Accordingly, in the representation shown, following approximately 21.2 seconds, the control is switched off.

[0025] In the following, the performance characteristics of the functioning monitoring are elucidated on the basis of these diagrams. In the top diagram, representing the actual state of the engine with respect to its running smoothness, the running smoothness up to instant t=5 seconds is optimal, since all cylinder-specific angular acceleration values are close to control setpoint value K=0. At instant t=5 seconds, a disturbance occurs in that less fuel is dispensed at the injector of the first cylinder 50%. As a result, the torque contribution of this cylinder or its angular acceleration value (curve K1) distances itself from the setpoint value and rises approximately within the next 5 seconds to a value of about 2.5. At the same time, the angular acceleration values of the remaining three cylinders (curves K2, K3, K4) gradually deviate by substantially the same amounts from the control setpoint value. The amount of the deviation is determined, inter alia, in that the control is designed, as a whole, such that the total fuel consumption of all cylinders essentially remains the same.

[0026] In the middle diagram, the reaction of the cylinder-specific controllers to the defect at the injector of the first cylinder is discernible. On the basis of curve R1, it becomes clear that the PI controller assigned to the first cylinder is attempting to compensate for the errors recognized by the control (50% less fuel at injector 1) by increasing the injection time. This is increased step-by-step in time with the controller (curve B1), in each instance in steps of 10% relative to the output value. At instant t=7.2 seconds, the increase in injection time amounts to 30%, which corresponds to the controller's preset intervention stop. Consequently, the limit of the controller's operating range is reached. On the basis of the controller setpoint input for a constant total injection quantity, the injection times or injection quantities of the other three cylinders (R2, R3, R4) are lowered in corresponding fashion step-by-step.

[0027] In response to intervention stop EA being reached at t=7.2, an error signal is output to the control where a first error bit B21 is set.

[0028] On the basis of curves K1 through K4 deviating substantially from K=0, it is discernible that, at the instant the intervention stop is reached, the running smoothness of the engine is not restored. Thus, the controller target point is not reached.

[0029] When the operating range limit of the control (intervention stop EA) is reached, this leads to changes in the controller characteristic of the PI controller assigned to the first cylinder. On the one hand, the controller is set back or reset, in particular to set the I component of the controller to zero again, so that preceding control deviations no longer influence the controller characteristics after it is reset. On the other hand, the controller loop gain is reduced, in the illustrated example, to half of the original controller loop gain existing at instant t=0. This means an increase in the controller ruggedness, with a simultaneous loss in controller's dynamic response. In other words, the controller becomes more sluggish and reacts more slowly to control deviations.

[0030] Since, in spite of the controller intervention between the occurrence of the disturbance (t=5 seconds) and the attainment of the intervention stop (t=7.2 seconds), there is no improvement in the running smoothness (curve K1 deviates with an increasing tendency from curves K2 through K4), the control again attempts in the time period following the first-time attainment of the invervention stop at T=7.2 seconds, to compensate for the signaled error (less fuel at the first injector) step-by-step in step with the controller by increasing the assigned injection time. After four increasing steps or a time interval of approximately 3.5 seconds, the controller again reaches its preset intervention stop at t=11 seconds, without the running smoothness disturbance being canceled. As a result, a second error signal (error bit B22) is generated, as well as in the manner already described, the control is again reset, and the controller loop gain is further reduced. This is reduced, for example, to one third of the original controller loop gain and, in this manner, the controller ruggedness is further enhanced.

[0031] It can be discerned that this process (resetting of the controller when the intervention stop is reached and a further reduction in the controller loop gain) is performed two more times in the illustrated example, in each case after the intervention stop is reached in the subsequent controller period, the time until the new intervention stop is reached increases, and correspondingly more steps for increasing the injection time are carried out.

[0032] It is discernible that even after the controller loop gain is canceled three times, the controller again reaches its intervention stop (t=21.2 seconds). In the illustrated specific embodiment of the control, this is evaluated as an indication that the attainment of the intervention stop is not attributed, for instance, to instabilities in the control (for example, due to parameter changes in the controlled system), but rather, actually to a cylinder defect, which, in the exemplary case, results in a considerable reduction in the injected fuel quantity at the first cylinder. The control is designed such that, following the fourth time the intervention stop is reached (t=21.2 seconds), the functioning of the control function is switched off by the functioning monitoring (setting of the function switch-off bits in curve B3). In this context, the intervention factors of all controllers are set back to value 1.

[0033] The last error bit or error signal that leads to the functioning monitoring being switched off and/or the switch-off signal at t=21.2 seconds in curve B3 may be used for activating an external warning indication, such as a warning lamp, and/or for producing an entry in a diagnostic memory, in order to show at a next engine diagnosis that a cylinder defect exists which is no longer able to be compensated by the cylinder equalization control. The error information may be specified to the extent that one is able to infer that this error occurred in the area of cylinder 1. Thus, targeted remedial measures, such as exchanging the injector of this cylinder, may be undertaken.

Claims

1. A method for controlling a multicylinder internal combustion engine, a cylinder equalization control being carried out for adjusting cylinder-specific torque contributions to the overall torque of the internal combustion engine, wherein the functioning of the cylinder-equalization control is monitored, and, in response to disturbances in the area of the cylinder-equalization control, at least one error signal is generated, indicating the existence of a disturbance.

2. The method as recited in claim 1, wherein in the context of the cylinder-equalization control, for each cylinder of the internal combustion engine, within a defined operating range, at least one combustion-influencing performance quantity is controlled on an individual basis, and an error signal is generated when at least one cylinder-specific control reaches a limit of the defined operating range.

3. The method as recited in claim 1 or 2, wherein in response to an error signal, a defined change in the control characteristic of the control follows, in particular by a canceling of the closed-loop gain.

4. The method as recited in claim 3, wherein subsequent to a first change in the control characteristic, as a response to a new error signal, the controller characteristic is changed at least one additional time, in particular through a subsequent cancellation of the control loop gain.

5. The method as recited in one of the preceding claims, wherein as a response to an error signal, preferably following an at least one-time change in the control characteristic, the control is switched off by the functioning monitoring, the control characteristic being reset, in particular, to the originally defined operating characteristics, or, as a response to an error signal, preferably following an at least one-time change in the control characteristic, the control is stopped by the functioning monitoring, the active controller characteristics being retained.

6. The method as recited in one of the preceding claims, wherein as a response to an error signal, preferably following an at least one-time change in the control characteristic, an external warning signal is emitted, in particular for activating a preferably optical warning display that is directly perceivable by the operator and/or for generating an entry into a memory of a diagnostic system capable of being externally queried.

7. A device for controlling a multicylinder internal combustion engine, comprising a cylinder equalization control for adjusting cylinder-specific torque contributions to the overall torque of the internal combustion engine, characterized by a device for monitoring the functioning of the cylinder-equalization control, which is designed such that, in response to disturbances in the area of the cylinder-equalization control, at least one error signal is generated, indicating the existence of a disturbance.