WO1999067525A1

WO1999067525A1 - Equalization of cylinder-specific torque contributions in a multi-cylinder internal combustion engine

Info

Publication number: WO1999067525A1
Application number: PCT/DE1999/001823
Authority: WO
Inventors: Stephan Uhl; Stefan Miller
Original assignee: Robert Bosch Gmbh
Priority date: 1998-06-25
Filing date: 1999-06-23
Publication date: 1999-12-29
Also published as: DE19828279A1

Abstract

The invention relates to an electronic control device for a quantity which influences irregular running of an internal combustion engine in a cylinder-specific manner. Said control device comprises a regulator for each cylinder to which a cylinder-specific irregular running signal is fed on the input side, and from which a cylinder-specific correction signal is output on the output side. The correction signal is combined with said quantity. The cylinder-specific correction signals are commonly processed in such a way that a cylinder-specific disturbance caused by correction interferences is counteracted in all cylinders, whereby the operating point shift in the cylinders not experiencing problems is compensated by another correction which globally detects all cylinders and which works in an opposed manner.

Description

WO 99/67525 _ - _ PCT / DE99 / 01823

Equalization of the cylinder-specific torque contributions in the multi-cylinder internal combustion engine

State of the art

The invention relates to the equality of the contributions of the individual cylinders to the total torque of an internal combustion engine. This can be, for example, an Otto or a diesel engine. The actual torque of a cylinder is recorded by evaluating the chronological course of the crank or

Camshaft rotation. A torque correction takes place via an intervention on at least one of the quantities of fuel injected, ignition timing in the gasoline engine,

Exhaust gas recirculation rate or injection position. The term injection position refers to the angular position of an injection pulse to a reference point, for example the top dead center of the piston of a cylinder in its combustion cycle.

A method for cylinder alignment is already known from EP 140 065. To evaluate the time course of the rotary movement of the crankshaft or camshaft, segment times are recorded in the known methods. Segment times are the times in which the crankshaft or camshaft covers a predetermined angular range that is assigned to a specific cylinder. The smoother the engine runs, the smaller the differences between the segment times of the individual cylinders. From the segment times mentioned can therefore be a measure of the uneven running of the engine. In the known methods, a control is assigned to each cylinder of the internal combustion engine, to which an actual cylinder uneven running value is fed as an input signal. The rough running values of several cylinders are averaged to form the control setpoint. The mean value serves as the setpoint. On the output side, the controller influences the cylinder-specific injection time and thus the cylinder-specific torque contribution in such a way that the cylinder-specific uneven running value approaches the setpoint.

The object of the invention is to further optimize cylinder equality. There is a particular need for a cylinder equalization function, particularly in the case of the direct-injection gasoline engine. This may be due to age-related changes in the flow characteristics of the high pressure injectors used in direct injection. With conventional gasoline engines with intake manifold injection, the smooth running of the engine can be improved, especially in lean operation.

This object of further optimization is solved with the features of claim 1. The invention advantageously uses the uneven running values which are formed in the control unit anyway for misfire detection. The formation of rough running values for misfire detection is known, for example, from US Pat. No. 5,861,553 (DE-OS 196 10 215). There will be as

Uneven running values LUT quotients are formed, in the numerator of which there are differences from successive segment times and whose denominator contains the third power of one of the segment times involved. This quotient can be weighted with other factors as well as with one Dynamic correction should be provided that takes into account changes in speed of the entire engine. With regard to the formation of uneven running values, the disclosure of the above-mentioned laid-open publication should be expressly included in this application. With the engine speed remaining the same and also with the speed dynamics, the sum of these rough running values formed by one camshaft revolution is zero. It is imperative to correct the mechanical errors of the segment time recording system (encoder wheel tolerance) during segment time recording in push mode, as these would otherwise be corrected. The consequence would be a real physical uneven running, which in connection with the mechanical errors would produce a signal of perfect smooth running. The aim of cylinder equalization is to minimize the real cylinder-specific angular accelerations, i.e. the rough running values, with a control concept. Then the torque components of the cylinders are the same.

Exemplary embodiments of the invention are explained below with reference to the figures. Fig. 1 shows the technical

Environment of the invention. Fig. 2 discloses a first one

Embodiment of the invention in

Function block diagram and FIG. 3 represents an embodiment of the invention supplemented by a further advantageous function.

1 in FIG. 1 represents a gasoline engine with direct injection, which is symbolized by a high-pressure injection valve 2, which projects into the combustion chamber 3 of the engine. 1 also shows a sensor wheel 4, sensors 5 and 5a and a control unit implemented as an electronic control device 6, which receives signals from the sensors 5 and 5a and outputs an injection pulse width tik to a cylinder-specific high-pressure injection valve. As is known, the injection pulse width is formed on the basis of further signals, for example via the amount of intake air, speed, temperatures, etc., which is known in detail to the person skilled in the art. The invention shown here relates to a correction of the injection pulse widths.

For this correction, cylinder-specific uneven running values are formed in the electronic control device and processed into correction values, which influence the torque and thus the temporal course of the rotation of the encoder wheel 4 via a calculation into the cylinder-specific injection times.

2 shows a structure of the function for forming the corrected injection times. The block 2.1

Signals 5 and 5a supplied. The signal 5a can, for example, indicate the top dead center in the combustion stroke of the first cylinder as the limit between two working cycles of a 4-stroke internal combustion engine. The signal 5 provides an angle information about the current angle of rotation of the

Master wheel with reference to a top dead center. From both information, block 2.1 determines segment times ts, which are assigned to the individual cylinders.

Blocks 2.2.1 to 2.2.z with z = number of cylinders calculate cylinder-specific uneven running values LUT from the segment times and blocks 2.3.1 to 2.3.z subject these uneven running values to filtering.

The cylinder-specific filtered uneven running values FLUT are compared with a nominal value, which can also be zero, which is symbolized by the number 2.4.

Alternatively, the uneven running values can also be made available to the controller unfiltered. The filtered Uneven running values are used for slow control interventions and the unfiltered uneven running values are used for fast interventions by the controller. Positive LUT values or FLUT values represent a deceleration, negative ones an acceleration of the crankshaft due to combustion in a cylinder. The amount of the LUT value (FLUT value) is the direct measure of the control deviation, which must be eliminated. A PI controller can be used for this.

Blocks 2.4 and 2.5 together represent a PI controller R1 for cylinder No. 1. The output variable of the controller is linked with a pilot control value from block 2.6 to form a manipulated variable R1.

In an analogous manner, further PI controllers R2 to Rz generate manipulated variables r2 to rz for all z cylinders of the internal combustion engine. These manipulated variables are fed to block 2.7.

Alternatively, control manipulated variables can also be formed using pattern recognition. Cylinders that give a lot or little torque can express five different typical behavior patterns in the four-cylinder:

1. All LUT values are small or zero. This means that all cylinders give the same torque.

2. A positive LUT value is followed by three negative LUT values. This means that the cylinder with the positive LUT value does not deliver enough torque.

3. Positive and negative LUT values alternate. This means that there are two cylinders giving less torque, which are offset in the firing order by 360 ° crankshaft angle. 4. Two positive LUT values are followed by two negative LUT values. This means that two cylinders give less torque, which are offset by 180 ° crankshaft angle in the firing order.

5. There are three positive and one negative values. This means that three cylinders give too little torque and one cylinder gives too much torque. This may be caused by a rich enrichment due to an injector fault in the cylinder that gives off too much torque.

Depending on the recognized pattern and the amount of the individual filtered and unfiltered LUT values, the following operations can be carried out on the engine individually or in combination:

1. Increase the injection time of the cylinders that give less torque, while reducing the injection time of the cylinders that give more torque.

The lengthening or shortening is calculated so that the total amount of fuel is not affected, i.e. that the lambda value of the mixture with which the engine is operated is not changed. This is the function of block 2.7 in

Fig. 2. This concept is used, for example, for lean control in the specified lambda value and maximum smoothness or minimal smoothness.

2. At the running limit, the difference between the current LUT value and the LUT mean value becomes significantly larger. The running limit is the uneven running, at which just stable burns take place, for example in homogeneous lean operation. The concept of homogeneity refers to the spatial variation of the Mixture composition in the combustion chamber: This is small in homogeneous operation. A distinction must be made between so-called stratified operation, in which the composition of the mixture in the combustion chamber is not homogeneous, but is, for example, more fuel-rich near the spark plug than at some distance from it. This running limit can be detected using the above method and regulated if necessary. With this concept, the total lambda changes, so it is not a matter of lean regulation, but of an individual cylinder

Limit control. Block 2.7 can be omitted in this implementation.

3. In today's systems, the ignition angles of an engine are stored in characteristic maps that apply to all cylinders. The regulation can therefore be designed so that an attempt is first made to compensate for a torque that is too low by means of an earlier ignition angle.

Example: Valve coking in the direct injection engine causes the cylinder torque to be too low. This cylinder runs leaner. In homogeneous operation, the optimal ignition angle for this lean mixture is earlier and can be adjusted accordingly. Likewise, cylinders that give off a lot of torque can be influenced by later firing angles.

4. In systems with exhaust gas recirculation, there is also the possibility of reducing the exhaust gas recirculation rate if the measured values fluctuate greatly. Here is one too

Limit control with the aim of maximizing the exhaust gas recirculation rate possible. 5. In the case of direct injection engines, the torque delivered can also be changed in shift operation by varying the injection position.

The main components of the function are the cylinder-specific PI controllers, which are identical in their structure and which regulate the filtered uneven running values FLUT buw LUT of the individual cylinders to zero. The PI control process only takes place in shift operation. In other words: the cylinder equalization function is only active in shift operation.

Block 2.7 has the following function: Standardization of the PI controller outputs ensures that the sum of the injection times of all cylinders remains constant and is not changed by numerical errors or errors caused by the process. Ideally, the buzzer of all I components of the controller is already zero, the resulting unequal component is evenly distributed to the other cylinders. The sum of all standardized controller outputs is zero and the sum of all injection times remains constant even with the controller interventions. An error bit is set if there is a controller intervention at the upper or lower stop.

The normalized control variables rln to rlz are then transferred to block 2.8, which represents the operating point adaptation that is essential to the invention.

The operating point is defined, among other things, by the quantity and composition of the cylinder filling. Example: The set of all possible (mixture quantity / lambda) points lie in a plane spanned by axes for lambda and mixture quantity. If the associated mixture composition Lambda is changed for a certain filling quantity, the position of this operating point changes. If, for example, a cylinder is excessively richly greased because, for example, the flow characteristics of its high-pressure injection valve differ greatly from the characteristics of the other injection valves, this has the consequence that the greasing of the problematic cylinder and the simultaneous emaciation of the other cylinders make the operating point particularly the cylinder with this problem is not moved. The operating point shift is corrected in the operating point adaptation essential to the invention: the operating point is tracked to the intact cylinders. The basis is the assumption that the behavior of the majority of the valves characterizes the intact cylinders, so that only operating point drifts can be adapted due to the failure of a single valve. The signing of each standardized controller output and its sum formation can be used to decide whether the operating point must be corrected with a global factor up or down. Global means in this

Context that the correction does not act individually for the cylinder, but covers all cylinders, that is to say acts globally. Since the operating point adaptation affects all cylinders globally, it has no influence on the behavior of the individual cylinder equalization PI controllers, but changes the injection time output by the defined working range. Updating this

The operating point adaptation factor is only slow and must be done much more slowly than the PI controllers.

The following example shows how the operating point adaptation works:

- Cylinder 1 converts 24% less fuel into torque, so the engine gives 6% less torque overall from. Before this effect is included in the control, the factors that have a direct effect on the injection time output have the following values -.

The four PI controllers now distribute the output injection times to the cylinders while maintaining the sum of all injection times until they are equal. Equality shows here that the different cylinders have the same rough running values. The following injection correction factor distribution then occurred:

As a result, the operating point of the intact cylinders 2, 3 and 4 has shifted by 6%, the defective cylinder has been restored to 6%.

The operating point adaptation now leads through global, i.e. for all cylinders uniform greasing by 6%, the intact cylinders back to the old operating point. The following factor distribution results:

If the malfunction disappears, cylinder 1 converts 24% more fuel into the torque, so that the PI controllers react with the following redistribution:

The operating point adaptation then leads all cylinders back to the correct operating point by global leaning by 6%:

This procedure has the following advantage: Without the operating point adaptation, the cylinder concerned is enriched when the flow of a valve is reduced. At the same time, the other intact cylinders are leaned by the same amount. This shift in the operating points of the intact cylinders can result in the operating points drifting from areas of stable combustion to areas of unstable combustion. This is counteracted by the operating point adaptation measure: It ensures stratified combustion stability when a single valve defect is detected by returning the operating point of the intact cylinders by means of global enrichment or leaning.

On the output side of block 2.8, the operating point-adapted and normalized control variables are linked with provisional values ti for injection pulse widths to cylinder-individually corrected injection pulse widths tik for the high-pressure injection valves 2a to 2c. The preliminary values for the injection pulse widths are provided by block 2.9.

3: The subject of FIG. 3 is based on the subject of FIG. 2 and supplements it. Each PI controller advantageously has a pilot control map 3.1, which is determined adaptively while driving. It reflects that cylinder-specific differences in the high-pressure injection valves and the uneven distribution in the pressure curve of the fuel rail. Depending on whether there is stratified or homogeneous operation, either the pilot control to relieve the PI controller and the dynamic improvement is included, or the factor determined from the pilot control indicators is used for the injection time correction for homogeneous operation. The controller output variable is constant over time in homogeneous operation. The PI controller does not operate in homogeneous mode, the cylinder equalization function is passive. The switchover is controlled by block 2.10 and carried out with switch 2.11. Switch 2.11 is closed in shift operation and opened in homogeneous operation.

The pilot control map 3.1 has X speed and Y load points. A 4-point interpolation is used to determine a pilot value for each speed / load point. The adaption of the pre-control indicators takes place with the adaption clock on average 3.2, which can be implemented as a low-pass filter, for example. The step size of the adaptation depends on the current difference between feedforward control and controller intervention. This difference is zero if the feedforward control corresponds exactly to the outputs of the PI controller.

As already mentioned at the beginning, the method is based on the fact that the sum of the rough running values over two crankshaft revolutions is zero. However, due to numerical errors and speed dynamics, this sum is slightly positive and therefore, in the long term, ie in periods longer than 15 minutes, the I components of the PI controllers of the individual cylinders slowly run together against the upper stop. This is avoided by a compensation integrator: by adding a method compensation value to the Actual values of all PI controllers are achieved depending on the status of the I components, ie their sum.

Claims

Expectations

1.Electronic control device for a variable that influences the uneven running of an internal combustion engine with individual cylinders, with one controller for each cylinder, to which a cylinder-specific uneven running signal is supplied on the input side and which outputs a cylinder-specific correction signal on the output side, which is linked to the specified size, the cylinder-specific ones Correction signals are processed together in such a way that a cylinder-specific malfunction due to correction interventions in all cylinders is counteracted, characterized in that the operating point shift in the non-problematic cylinders is compensated for by a counter-acting further correction which affects all cylinders globally.