DE10016886A1 - Method and device for regulating an internal combustion engine - Google Patents

Abstract

The method involves measuring the residual oxygen in the exhaust gas with a linear broad band lambda probe and passing a corresponding signal to a controller that generates a signal for influencing the air-fuel ratio into the engine. The actual lambda probe characteristic gradient is determined while it is in use for controlling an engine and its deviation from a desired gradient determined, from which a probe signal correction value is derived. An Independent claim is also included for an arrangement for lambda control of an internal combustion engine.

Description

The invention relates to a method for controlling an internal combustion engine the features mentioned in the preamble of claim 1 and a device with the features mentioned in the preamble of claim 8.

To control an operating mode of an internal combustion engine, it is known in to arrange at least one lambda probe in an exhaust gas duct. With the help of Lambda probe, a residual oxygen concentration in the exhaust gas can be detected and from this a conclusion on the ratio of an oxygen fraction to one Fuel fraction in an air-fuel mixture supplied to the combustion process respectively.

A signal provided by the lambda probe is fed to an engine control unit, this results in an influence on the composition of the fuel-air mixture Control signal generated. Using this control signal, for example Fuel injection or an air supply to the Internal combustion engine to be regulated.

So-called two-point lambda probes are known, which are highly non-linear Have a characteristic curve with a very steep transition at λ = 1. By means of such Lambda sensors can only detect two states, namely there is fuel in the stoichiometric excess or air is in the stoichiometric excess of Air-fuel mixture before.

In order to achieve improved lambda control of the internal combustion engine, in particular for different operating modes Internal combustion engine, such as cold start or full load, respectively for different types of internal combustion engines, such as Stratified or lean-burn engines, the use of so-called broadband Lambda sensors known. These broadband lambda sensors are characterized by a  linear characteristic curve, which over a large range, for example λ = 0.7 to 5, runs. With such linear broadband lambda probes, a continuous Lambda control are carried out, the setpoint of which in the entire measuring range linear broadband lambda probe. Such linear broadband Lambda sensors are also characterized by a faster response and a more precise regulation.

When using such linear broadband lambda probes, however, it is disadvantageous that these, due to manufacturing tolerances, aging influences, poisoning influences or the like, have an offset or a slope deviation in the characteristic can affect the precise regulation. Especially at a relatively far from λ = 1 lying control values for lambda an error band (tolerance band) widens linear characteristic. The course of the linear characteristic curve is particularly lean Fuel-air mixtures with λ <1 relatively flat, so that there are deviations in the Characteristic curve lead to relatively large errors. Such errors lead to the fact that an incorrect lambda is adjusted in relation to the desired lambda. Out of this results in an impairment of the combustion quality, a higher one Pollutant emissions and increased fuel consumption.

DE 196 29 552 C1 describes a device for compensating for the temperature drift a linear broadband lambda probe known. Here it is suggested in one Control unit to store a map that depends on the working temperature of the Lambda probe contains associated values for the output signal. This will possible, depending on the actual temperature, a temperature drift of the lambda sensor compensate.

DE 195 45 706 A1 describes a method for calibrating a two-point Lambda sensor known in an internal combustion engine, in which a catalyst for a certain period of time with an over-rich fuel-air mixture is supplied and the signal values of the lambda probe during this period can be measured independently of other control signals. This is supposed to Correction value are formed during normal operation of the internal combustion engine is fed to the probe signal.  

EP 0 686 232 B1 discloses a linear broadband lambda probe with a Combine two-point lambda probe to create the linear broadband lambda probe to calibrate a value λ = 1 using the two-point lambda probe.

US Pat. No. 5,473,889 describes an arrangement in which, on the one hand, a linear Broadband lambda probe and, on the other hand, a further gas measuring probe are provided are. Based on the measurement results provided by the two probes, the linear broadband lambda probe signal can be verified.

Finally, EP 0 894 187 B1 describes a method for model-based Transient control of an internal combustion engine is known in which Fuel deposits in the intake manifold using a so-called wall film model are taken into account, whereby the parameters of the wall film model include an output signal of one in the exhaust tract of the internal combustion engine arranged linear broadband lambda probe can be adapted.

So-called regression calculations are known from general measurement technology, by means of which an approximation curve from a given number of measuring points can be determined. In the result of the regression calculation, parameters are the Approximation curve, for example a slope of a straight line or a Offset, representable.

The invention has for its object a method and an apparatus Specify generic type, by means of which an accurate Lambda control of internal combustion engines in a wide control range is feasible.

According to the invention, this object is achieved by a method with the method described in claim 1 Features mentioned and a device with those mentioned in claim 8 Features resolved. Because an increase in the actual characteristic of the linear broadband Lambda sensor - during the intended use of the linear broadband Lambda sensor in the control of an internal combustion engine - is determined, a Deviation of the slope from a target characteristic curve is determined and from the deviation a correction value is determined with which the control signal for influencing the Composition of the fuel-air mixture for operating the Correcting the internal combustion engine is advantageously possible, automatically continuously  adapt the actual characteristic of the linear broadband lambda probe. So you can in particular without additional component expenditure, for example by additional Probes or the like, for calibrating the linear broadband lambda probe exact Lambda values can be specified. This also advantageously makes it possible for Manufacture of linear broadband lambda probes greater manufacturing tolerances allow, because during the intended use of the lambda sensors a the existing errors are automatically compensated. This is what happens a reduction in production costs due to quality monitoring requirements etc. can be reduced and on the other hand the yield of Manufacturing process is increased.

Furthermore, it is advantageously possible for the method according to the invention to be age-related or to compensate for changes in the characteristic curves caused by poisoning that the service life of the linear broadband lambda sensors can be increased.

In a preferred embodiment of the invention it is provided that the slope of the actual Characteristic is determined by a regression calculation, in which preferably the Regression calculation the output signals of the linear broadband lambda probe and evaluates a lambda target signal for operating the internal combustion engine. By Such a linkage of the signals can be used to determine a regression factor, which is the represents the average slope of the actual characteristic curve at the point of the lambda target mean value. This regression factor can preferably be used with the desired characteristic curve linear broadband lambda probe and the desired lambda value are linked, so that a correction factor is available by means of which the linear Broadband lambda sensor supplied signal can be corrected. This will create a exact lambda control of the internal combustion engine, in particular also via a wide lambda control range, in rich and lean operating modes Internal combustion engine possible.

Furthermore, the method according to the invention enables an onboard Diagnosis possible because the regression means the deviation of the actual characteristic the linear broadband lambda probe can be determined from the ideal characteristic curve. This Deviation of the actual from the ideal characteristic can be done in a simple manner a maximum allowable deviation are compared so that when the maximum permissible deviations to an error case is detected, for example an exchange of the linear broadband lambda probe requires.  

Further preferred refinements of the invention result from the others in the Characteristics mentioned subclaims.

The invention is described in an exemplary embodiment based on the associated Drawings explained in more detail. Show it:

Fig. 1 shows schematically a equipped with a lambda control the internal combustion engine;

FIG. 2 shows two curves of the linear wide-band lambda probe;

Fig. 3 is a block diagram of the lambda probe correction according to the invention and

Fig. 4 shows the profile of individual characteristics of the oxygen sensor correcting the invention.

Fig. 1 shows schematically an internal combustion engine 10, the exhaust pipe 12 is connected to a catalytic converter, in particular a 3-way catalytic converter 14. A linear broadband lambda probe 16 (hereinafter lambda probe 16 ) is arranged in the exhaust line 12 . A signal line 18 of the lambda probe 16 is connected to an engine control unit 20 .

The internal combustion engine 10 further comprises an intake line 22 , in which a means 24 for adjusting an amount of intake air is arranged. Furthermore, a means 26 for introducing, for example injecting, a fuel into the internal combustion engine 10 , in particular into the intake air, is provided. The means 24 and 26 are also connected to the engine control unit 20 via control lines 28 and 30, respectively. The engine control unit 20 has further connections, which are only indicated here, with which a multiplicity of monitoring, control, regulation or the like of the internal combustion engine 10 can be carried out. However, this will not be discussed in more detail in the context of the present description.

The general mode of operation of the schematic arrangement shown in FIG. 1 is as follows:

A fuel-air mixture is burned in the internal combustion engine 10 in order to generate drive energy, for example for a motor vehicle. The exhaust gas from the combustion process is fed via the exhaust gas line 12 to a catalytic converter 14 , by means of which nitrogen oxides NO _x , hydrocarbons HC or carbon monoxide CO are absorbed, for example. The exhaust gas is guided past the lambda probe 16 , by means of which a residual oxygen content of the exhaust gas 12 can be measured in a known manner. A signal corresponding to the residual oxygen content is transmitted from the lambda probe 16 to the engine control unit 20 . The engine control unit 20 provides control signals for the means 24 and 26 , by means of which an air quantity and / or a fuel quantity for the fuel-air mixture to be burned in the internal combustion engine 10 is metered. The signal provided by the lambda probe 16 is taken into account here in accordance with a predetermined lambda value.

In Fig. 2 is a characteristic curve of the linear wide-band lambda probe designated 16 32. The output voltage U _{LSU of} the lambda probe 16 is plotted against the lambda value. Conventional lambda probes 16 have an output voltage U of 2.5 V when the lambda value = 1. For comparison, a second characteristic curve 34 is entered, which has errors. This has an overall lower gradient around the value λ = 1. That is, this faulty characteristic curve 34 has in the lean region (λ <1) to a low output voltage U _LSU and in the rich region too high an output voltage (λ <1) U _LSU. This leads to a gradient .DELTA.U _LSU 34 being smaller than a gradient .DELTA.U _LSU 32 by the value λ = 1, since the characteristic curve 34 has a smaller gradient by the value λ = 1.

Such faulty characteristic curves 34 can occur, for example, as a result of manufacturing tolerances, signs of aging or signs of poisoning. In any case, this leads to a deviation of the slope of the actual characteristic curve 34 from the target characteristic curve 32 .

FIG. 3 shows in a block diagram the lambda probe correction according to the invention, by means of which, when the lambda probe 16 ( FIG. 1) is used as intended, the offset of the characteristic curve 34, for example explained with reference to FIG. 2, can be compensated for by the desired characteristic curve 32 .

The individual components of the lambda sensor correction are integrated in the engine control unit 20 .

The voltage signal U _{LSU of} the lambda probe 16 is supplied on the one hand to a regression calculation 36 and on the other hand to a subtractor 38 . The regression calculation 36 is also supplied with a signal λ _target . The signal λ _target is also fed to a storage means 40 in which the target characteristic curve 32 is stored. On the basis of the characteristic curve 32 stored in the storage means 40 , a signal U _{LSU model is} determined from the signal λ _{target and is} supplied to a differentiator 42 . The differentiator 42 is simultaneously subjected to the signal λ _target . At the output of the differentiator 42 there is a signal dU _LSU-model / dλ _Soll which corresponds to the nominal slope of the characteristic curve 32 .

At the output of the regression calculation 36 there is a signal dU _LSU / dλ _Soll which corresponds to the actual slope of the characteristic curve 34 . This signal corresponds to the regression factor R _{LSU of} the lambda probe 16 . This represents the average slope of the actual characteristic curve 34 at the point of the lambda target mean value. The regression factor R _{LSU is} linked to the nominal slope of the characteristic curve 32 via a ratio element 44 and fed to an adaptation element 46 . The adaptation element 46 determines a correction factor K _LSU for the lambda probe 16 from the input signal.

The actual signal U _{LSU of} the lambda probe 16 is linked via the subtractor 38 with a voltage of 2.5 V, which corresponds to the target voltage at λ = 1. The resulting difference is linked to the correction factor LSU via a multiplier 48 . The multiplier 48 is connected to a summing element 50, over which the previously drawn in the subtracter 38 voltage value 2, 5 V is added to the signal again, so that as the output signal a corrected voltage signal U _{LSU korrigtert} the lambda probe 16 is available and by the engine control unit 20 ( Fig. 1) can be used for the control of the internal combustion engine 10 .

It is clear that the signal processing means that a deviation of the slope of the actual characteristic curve 34 from the slope of the target characteristic curve 32 can be corrected in a simple manner.

FIG. 4 shows typical signal profiles of the lambda sensor correction explained with reference to FIG. 3. The individual signal curves are plotted against time t. 60 is a target value for the gradient of the desired characteristic curve 32, so U _LSU d _model / dλ _target shown. A curve of the regression factor R _{LSU is also} shown at 62. A characteristic curve is plotted at 64 , which represents noise of the voltage signal U _LSU . The characteristic curve 64 oscillates here by a noise factor RF of 1.

The characteristic curve of the voltage signal U _LSU corresponding to the desired characteristic curve 32 ( FIG. 2) is plotted at 66 for a lambda control of ± 3%. This means that the value λ _target fluctuates (toggles) by a value of 0.97 to 1.03. The signal disturbances actually present by the noise signals 64 lead to a characteristic curve 68 , which results from a superimposition of the characteristic curve 66 with the characteristic curve 64 . The characteristic curve 66 thus corresponds to the desired signal, while the characteristic curve 68 corresponds to the actual signal. On the basis of the characteristic curve 62 , it becomes clear that the regression factor R _LSU approaches the characteristic curve 60 , which corresponds to the nominal slope of the characteristic curve 32 , after a short time. The regression factor 62 is recalculated for all newly added measured value pairs, this is the actual measured voltage U _{LSU of} the lambda probe 16 and the value λ _target at any time, over all measured values of the measurement cycle. The measurement cycle begins, for example, each time the internal combustion engine 10 is restarted. As a result, the number of measured values included in the regression increases with each new pair of measured values. It becomes clear that after a short time, in particular within a few seconds, the course 62 of the regression factor approaches the course of the characteristic curve 60 .

On the basis of this model representation of the characteristic curves, it is clear that, despite the superposition of the lambda target signal (characteristic curve 66 ) by the noise signal 64, the output signal U _LSU corrects ( FIG. 3) by continuously determining the regression factor R _LSU in the lambda control of the internal combustion engine 10, the deviation of the gradient the actual characteristic curve 34 is compensated for by the target characteristic curve 32 .

Based on the explanation, it is also clear that the lambda probe correction, as shown in FIG. 3, allows the slope of the actual characteristic curve 34 to deviate from the target characteristic curve 32 for each linear broadband lambda probe 16 . This deviation can be caused by manufacturing tolerances, aging or signs of poisoning. This is not a criterion for the correction of the lambda probe voltage signal U _LSU .

The on-board diagnosis of the motor vehicle having the internal combustion engine 10 can be carried out simultaneously by the lambda sensor correction. If the regression factor R _LSU and / or the correction factor K _{LSU is} so large that the deviation of the actual characteristic curve 34 from the target characteristic curve 32 exceeds a predeterminable maximum deviation, an error of the lambda probe 16 which can no longer be caused by the inventive method can be derived from this Correction can be compensated, be closed. The exchange of the corresponding lambda probe 16 can be displayed by providing a corresponding signal.

REFERENCE SIGN LIST

10th

Internal combustion engine

12th

Exhaust pipe

14

3-way catalytic converter

16

Broadband lambda sensor

18th

Signal line

20th

Engine control unit

22

Suction pipe

24th

medium

26

medium

28

Control lines

30th

Control lines

32

Target characteristic

34

Actual characteristic

36

Regression calculation

38

Subtractor

40

Storage means

42

Differentiator

44

Relationship

46

Adapter

48

Multiplier

50

Summer

60

Setpoint for the slope of the set characteristic

32

62

Course of the regression factor Risu

64

Noise signals

66

Characteristic curve

68

Characteristic curve
U _LSU

Lambda probe voltage signal
R _LSU

Regression factor
K _LSU

Correction factor
RF noise factor

Claims (8)

1. Method for regulating an internal combustion engine, wherein the residual oxygen content of an exhaust gas of the internal combustion engine is measured by means of a linear broadband lambda probe and a signal dependent on the residual oxygen content of the exhaust gas is transmitted to a control device and the control device uses this to supply the composition of the internal combustion engine with fuel-air. Control signal influencing the mixture is generated, characterized in that a slope of the actual characteristic curve of the linear broadband lambda probe is determined - during the intended use of the linear broadband lambda probe in the control of an internal combustion engine - a deviation of the slope from a target characteristic curve is determined and a correction value is determined from the deviation, with which the signal of the linear broadband lambda probe is corrected. 2. The method according to claim 1, characterized in that an increase in the actual Characteristic curve is determined by a regression calculation. 3. The method according to any one of the preceding claims, characterized in that the regression calculation, the actual signal (U _LSU ) of the linear broadband lambda probe and a setpoint (λ _set ) of the lambda control is supplied. 4. The method according to any one of the preceding claims, characterized in that the slope of the target characteristic curve is determined from a known target characteristic curve and the target value (λ _target ). 5. The method according to any one of the preceding claims, characterized in that a regression factor (R _LSU ) of the regression calculation is linked to the slope of the target characteristic curve and from this a correction factor (K _LSU ) for the actual signal (U _LSU ) is determined. 6. The method according to any one of the preceding claims, characterized in that the regression factor (R _LSU ) is determined in predetermined time steps with each newly added pair of measured values of the actual signal (U _LSU ) and the target value (λ _target ) over all measured values of the measurement cycle . 7. The method according to any one of the preceding claims, characterized in that the regression factor (R _LSU ) and / or the correction factor (K _LSU ) is used for on-board diagnosis of a motor vehicle having the internal combustion engine. 8. Device for lambda control of an internal combustion engine, which comprises at least one linear broadband lambda probe arranged in an exhaust gas line, and has a control unit by means of which a signal supplied by the lambda probe can be detected and the means for creating a fuel-air mixture for the internal combustion engine actuated, characterized by means ( 36 ), by means of which an incline of an actual characteristic curve ( 34 ) of the lambda probe ( 16 ) can be determined and means ( 40 , 42 ), by means of which an incline of a desired characteristic curve ( 32 ) can be ascertained, and means ( 44 , 46 ) for determining a correction value (K _LSU ) of the signal (U _LSU ) of the lambda probe ( 16 ).