Method and device for monitoring gas sensors

Abstract

A method and a device for monitoring gas sensors in an internal combustion engine, in which in a steady-state operation of the internal combustion engine, the output signal of the gas sensor is filtered by a high-pass filter and higher-frequency signal components are analyzed by a comparison with an appropriately processed model value. When using the method and the device for executing the method, electrically oscillating gas sensors or the incoupling of interference variables or faults in the evaluation circuit, especially in the case of exhaust gas sensors in an exhaust-gas purification and reducing system, are able to be detected, which minimizes faulty interpretations in a dynamic diagnosis. This monitoring function is advantageously combinable with dynamic diagnosis functions that likewise analyze higher-frequency signal components by a comparison with appropriately processed model values. This increases the operational reliability of the exhaust-gas purification or reducing system.

Claims

1. A method for diagnosing a gas sensor of an internal combustion engine, the method comprising: comparing a modeled signal and a measured signal of a gas state variable to provide a comparison thereof; performing a dynamic diagnosis in response to a change in the gas state variable based on the comparison of the modeled signal and the measured signal, the measured signal being an actual value of an output signal of the gas sensor, and the modeled signal being a model value based on input variables of the internal combustion engine, wherein in a steady-state operation of the internal combustion engine, the output signal of the gas sensor is filtered by a high-pass filter, wherein the dynamic diagnosis includes analyzing a signal energy of higher-frequency signal components of the measured signal by comparing the signal energy with a signal energy of a processed model value; wherein the gas sensor exhibits a low-pass behavior as a function of a geometry of a protective sleeve, a measuring principle, an aging or a contamination, wherein the high-pass filter is connected in series with the gas sensor, which exhibits a low-pass behavior, and for steep load changes, an output signal of the high-pass filter is used to detect whether a limit frequency of the low-pass behavior of the gas sensor is greater or smaller than another limit frequency of the high-pass filter, and wherein the dynamic diagnosis includes inferring an electrical fault of the gas sensor or an occurrence of a vibration based on the comparison.

2. The method of claim 1, wherein a rate of change of the gas state variables to be measured is ascertained and is used to detect a steady-state operation of the internal combustion engine.

3. The method of claim 1, wherein the energy or the power of the higher-frequency signal components of the gas sensor and of correspondingly high-pass filtered output signals from a model of the gas sensor is compared to threshold values for the energy or the power, and the presence of electrical faults of the gas sensor or the occurrence of oscillations is inferred based on the comparison.

4. The method of claim 3, wherein when an upper threshold value for the energy or the power of the higher-frequency signal components of the gas sensor is exceeded, and a simultaneous drop occurs below a lower threshold value for the energy or the power of the higher-frequency signal components of the modeled signal determined from the model, an electrically oscillating gas sensor is inferred.

5. The method of claim 1, wherein a model time constant stored in the model corresponds to that of a nominal gas sensor and/or this model time constant as well as the threshold values are adapted as a function of the gas state variables.

6. The method of claim 1, wherein the integration of the higher-frequency signal components of the sensor signal and the modeled signal is event-controlled or, if a steady-state operation of the internal combustion engine is detected, is started and terminated in a time-controlled or event-controlled manner.

7. The method of claim 6, wherein in case of an event control, the threshold values are adapted as a function of an instantaneous integration period, which deviates from a typical integration period.

8. The method of claim 1, wherein an average signal power of the higher-frequency signal components of the sensor signal and the modeled signal is analyzed, and the comparison takes place with an applicable power threshold value.

9. The method of claim 1, wherein in the high-pass filtering the signal of the gas sensor and/or of the modeled signal is additionally filtered using filter units or filter functions that exhibit insensitivity or dead zones in their characteristic curve in the range of small input variables, which may feature step changes in their characteristic curve in addition.

10. The method of claim 9, wherein the characteristic curves of these filter units or filter functions are combined with the functionality of squaring the signals, or the absolute amount of the high-pass signal output is used or the absolute value generation is combined with the insensitivity or dead zone in one characteristic curve of the filter units.

11. The method of claim 1, wherein, as gas sensors, gas pressure sensors, gas temperature sensors, gas mass flow sensors or gas concentration sensors are used as exhaust gas sensors in the exhaust duct of the internal combustion engine, as part of an exhaust-gas monitoring and reducing system, or in an air supply of the internal combustion engine.

12. The method of claim 1, wherein exhaust gas sensors in the form of broadband lambda sensors or NO.sub.x sensors are used as gas sensors to ascertain an oxygen concentration in a gas mixture.

13. The method of claim 1, further comprising: using a process for dynamically diagnosing gas sensors, in which an analysis of higher-frequency signal components in a change of the gas state variable to be measured is performed for the dynamic diagnosis.

14. A device for diagnosing a gas sensor in the exhaust duct of an internal combustion engine as part of an exhaust gas monitoring and reducing system or in an air supply of the internal combustion engine, comprising: a diagnosing arrangement, in a monitoring unit, to perform a diagnosis in response to a change in a gas state variable to be measured, based on a comparison of a modeled signal and a measured signal, the measured signal being an actual value of an output signal of the gas sensor, and the modeled signal being a model value based on input variables of the internal combustion engine, wherein the monitoring unit has high-pass filters for analyzing higher-frequency signal components of the measured signal and at least one model value for the gas sensor, wherein the diagnosing arrangement is configured to analyze a signal energy of higher-frequency signal components by a comparison with a signal energy of an appropriately processed model value; wherein the gas sensor exhibits a low-pass behavior as a function of a geometry of a protective sleeve, a measuring principle, an aging or a contamination, wherein the high-pass filter is connected in series with the gas sensor, which exhibits a low-pass behavior, and for steep load changes, an output signal of the high-pass filter is used to detect whether a limit frequency of the low-pass behavior of the gas sensor is greater or smaller than another limit frequency of the high-pass filter, and wherein the diagnosing arrangement infers an electrical fault of the gas sensor or an occurrence of a vibration based on the comparison.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 shows in a schematic representation, the technical environment in which the method according to the present invention is able to be used.

(2) FIG. 2 shows a block diagram of a dynamic diagnosis circuit including a comparison of a signal energy, which can be used as the basis of a plausibility check for detecting oscillations or in coupled electrical interference variables.

(3) FIGS. 3a and 3b show a high-pass filter as part of the dynamic diagnosis circuit including a filter unit which has an insensitivity region without (FIG. 3a) and with a step change (FIG. 3b).

(4) FIGS. 4a and 4b show two alternative developments of the insensitivity region without a step change.

(5) FIGS. 5a and 5b show two alternative developments of the insensitivity region with a step change.

(6) FIG. 6 show a block diagram of a dynamic diagnosis circuit, as an alternative to the one in FIG. 2, featuring a comparison of an average signal power.

DETAILED DESCRIPTION

(7) Using an Otto engine as example, FIG. 1 schematically shows the technical environment in which the method for diagnosing an exhaust gas sensor 15 according to the invention is able to be used. An internal combustion engine 10 is supplied with air via an air supply 11, its mass being determined by an air mass meter 12. Air mass meter 12 may be configured as a hot-film air-mass meter, for example. The exhaust gas of internal combustion engine 10 is discharged via an exhaust gas duct 18, and an exhaust gas purification system 16 is provided downstream from internal combustion engine 10 in the flow direction of the exhaust gas. Exhaust gas purification system 16 usually includes at least one catalyst.

(8) An engine controller 14 is provided for the control of internal combustion engine 10, which first of all, supplies fuel to internal combustion engine 10 via a fuel metering device 13 and, secondly, receives the signals of air mass meter 12 and exhaust gas sensor 15 disposed in exhaust duct 18 and an exhaust gas sensor 17 situated in exhaust discharge line 18. In the example illustrated, exhaust gas sensor 15 determines an actual lambda value of a fuel-air mixture supplied to internal combustion engine 10. It may be developed as a broadband lambda sensor or as a continuous lambda sensor. Exhaust gas sensor 17 determines the exhaust gas composition downstream from exhaust gas purification system 16. Exhaust-gas sensor 17 may be developed as a step-change sensor or as a binary sensor.

(9) For better monitoring of the dynamics of exhaust gas sensor 15, it may be provided to use high- and low-pass filters in order to check whether exhaust gas sensor 15 is still able to detect the higher-frequency components of a concentration change in a load change of internal combustion engine 10. Such gas sensors have a characteristic low pass behavior, which depends on the geometry of their protective sleeve, among other things. In Diesel engines, such a protective sleeve may be contaminated with soot, so that the bandwidth of the sensor decreases. In the time range, the decreasing limit frequency manifests itself in a greater rise time, i.e., the signal flanks become flatter in response to the same excitation. If a suitable high-pass filter is therefore connected in series with the sensor, it is possible in steep load changes to detect from the output signal of the high pass whether the limit frequency of the low pass is greater or smaller than the limit frequency of the high pass. The dynamic response of the sensor may be inferred by analyzing these higher-frequency signal energies, as described in German patent document 10 2011 088 296.0, filed on Dec. 12, 2011.

(10) If the signal energy of an exhaust gas sensor 15 assumes an implausibly high value following the high-pass filtering in the steady-state operation, it is furthermore possible to infer an electrical oscillation or an incoupling of interference signals. A model may be used as reference for the plausibility check, as described in German patent document 10 2011 088 296.0, filed on Dec. 12, 2011.

(11) The present invention utilizes a filter system from German patent document 10 2011 088 296.0, filed on Dec. 12, 2011 for searching for high-frequency components in the sensor signal that should actually not be present, during steady-state engine operating phases. If a suitable high-pass filter is connected in series with exhaust gas sensor 15, the steady component and low-frequency components of the measuring signal are suppressed. Only the measuring noise may therefore contribute to the output power of the high-pass filter in a steady-state engine operation.

(12) FIG. 2 shows the basic structure of the dynamics monitoring from German patent document 10 2011 088 296.0, filed on Dec. 12, 2011 in the form of a block diagram 20. Shown in the upper portion is a path for an oxygen concentration 21 measured using exhaust gas sensor 15. Because of a real gas propagation time and sensor delay 22, which are able to be described by a dead time T.sub.t or a first order low-pass filter having a sensor time constant T.sub.S, an oxygen sensor signal 22.1 results on the basis of actual oxygen concentration 21. The transfer function of the sensors and gas propagation time 22 is obtained from the following relationship, in which K.sub.S represents an amplification factor for the sensor:
G(j)=K.sub.S exp(T.sub.t j)/(T.sub.S j+1)(2)

(13) K.sub.S normally corresponds to the multiplicative or rise error of the sensor that stems from production variances and aging. However, if it is not the oxygen concentration that is used as the sensor signal, but a variable that is proportional to said concentration, then K.sub.S is a corresponding transfer coefficient for converting the sensor signal into an oxygen concentration and may also be dimensional. Afterwards, oxygen sensor signal 22.1 is filtered by a high pass 23 and squared using a multiplier 24, thereby obtaining a signal that corresponds to a signal power. An integrator 25 subsequently integrates this signal, so that a signal energy 25.1 of the higher-frequency energy components of the measured oxygen content is obtained. A comparison with a correspondingly conditioned signal for a value ascertained with the aid of a model in a downstream evaluation unit 26 results in a status value 26.1, which may be used for the diagnosis.

(14) Because of incouplings on the cable harness or electrical faults of the evaluation circuit, for example, electrical interference variables 34 may be coupled into the sensor path, as shown in FIG. 2. The low pass behavior of the sensor naturally has no effect on the spectrum of the interference variables. Accordingly, sooting of the sensor, for example, has no effect on the fault sensitivity of the sensor path either.

(15) In the most basic case, high pass 23 may be developed as a first order high pass, whose transfer function is able to be described by the relation
G(j)=T.sub.F j/(T.sub.F j+1)(3)

(16) using T.sub.F as limit frequency of the filter. If the limit frequency of exhaust gas sensor 15 exceeds limit frequency T.sub.F of high pass 23, then the series connection behaves like a band pass, i.e., the high frequencies of the input spectrum of exhaust gas sensor 15 are still permitted to pass and may be detected in the output spectrum. In contrast, if the limit frequency of exhaust gas sensor 15 drops below limit frequency T.sub.F of high pass 23 due to a loss in dynamics, then the series circuit blocks all frequencies, so that it is no longer possible to measure any frequency components at all in the output spectrum.

(17) In principle, the present invention is not restricted to first order high pass filters. Instead, it is also possible to use any other high pass filters. In the same way, the monitoring method is usable when the low pass filters including exhaust gas sensor 15 itself are parameterized in a different manner, e.g., using a limit frequency instead of the time constant, or when they are of a higher order.

(18) To make it possible to distinguish between a slow exhaust gas sensor 15 and insufficient excitation when using the method from German patent document 10 2011 088 296.0, filed on Dec. 12, 2011, the rate of change of the exhaust gas composition must be evaluated, which in the case of a broadband lambda sensor, for example, may be accomplished on the basis of a change in the air and fuel mass. A similar series connection of filters may be used for this purpose. In the case of a broadband lambda sensor, this will require no more than converting the above masses into an O.sub.2 concentration and a delay using a low pass filter which corresponds to a functional exhaust gas sensor. This low pass filter must then be switched in series with a high pass that features the same transfer function as the real sensor. By comparing the two high pass outputs, it will then be possible to infer the operativeness of the real sensor. In case of another gas component, it may become necessary to use an additional untreated emissions model.

(19) The processing of the energy value determined with the aid of a model is illustrated in the lower part of block diagram 20 in FIG. 2. Following a stoichiometric correction, a quotient is formed in a division unit 29 from an air mass 27.sub.mL and a setpoint fuel mass 28 m.sub.K for fuel metering 13, and a lambda value is calculated. Fuel mass 28 may result from the torque request specified by the driver, which is converted into a fuel quantity. In a conversion unit 30, the lambda value is used to determine a calculated oxygen content 30.1. According to a model 31, the transfer function
G(j)=exp(T.sub.tM j)/(T.sub.M j+1)(4)

(20) is used to calculate a modeled oxygen content 31.1, T.sub.tM representing a model dead time, and T.sub.M representing a model time constant.

(21) Modeled oxygen content 31.1 then is filtered by another high pass 23, whose transfer function in the most basic case corresponds to that of the first order high pass, and squared by a further multiplier 24, which results in a signal that corresponds to a signal power. This signal is subsequently integrated by another integrator 25, so that signal energy 25.1 is obtained for the higher-frequency energy components of the modeled oxygen content.

(22) Since high pass 23 suppresses the steady component and the lower-frequency components, only the higher-frequency components of the individual O.sub.2 signal 22.1, 31.1 render a contribution. In the steady-state operation, the two high pass output signals Y.sub.S for the sensor signal and Y.sub.M for the model signal should therefore vanish, if the noise is disregarded. The two signal energies 25.1
.sub.M=.sub.0.sup.TY.sub.M.sup.2(t)dt(5a)
.sub.S=.sub.0.sup.TY.sub.S.sup.2(t)dt(5b)

(23) should consequently likewise assume very low values in the steady-state operation, T representing the integration period.

(24) The comparison of the two signal energies 25.1 in evaluation unit 26 now makes it possible to infer an electrical fault of exhaust gas sensor 15. If energy .sub.M of the model path is lower than a lower threshold .sub.unten, and energy .sub.S of the sensor path is simultaneously greater than an upper threshold .sub.oben, this may be interpreted to mean that the engine operating point is approximately constant and the sensor signal fluctuates nevertheless, which points to an electrically oscillating exhaust gas sensor 15. In summary, the following applies:
.sub.M<.sub.unten und .sub.S>.sub.oben.fwdarw.sensor is oscillating

(25) To improve the selectivity of the diagnosis, it is advisable to use a so-called insensitivity or dead zone, as is common practice when filtering noise. This suppresses small values of its input variable in an applicable range. This insensitivity zone may be realized by additional filter units 32, 33 shown in FIGS. 3a, 3b, 4a, 4b as well as 5a and 5b or by filter functions with their characteristic curves, for example, which may also feature a step change (filter unit 33 in each case).

(26) FIG. 3a shows the sensor path in a cutaway view. Oxygen sensor signal 22.1 is filtered by high pass 23, which has a filter unit featuring an insensitivity region (dead zone) without step change 32. The signal filtered in this manner is then squared by multiplier 24 and integrated using integrator 25, so that signal energy 25.1 is obtained on the output side.

(27) Small signal amplitudes about the zero point are suppressed in this case, as is obvious from the characteristic curve.

(28) FIG. 3b shows a system as an alternative to that of FIG. 3a; here, a filter unit 33 having an insensitivity region with a step change is used, which also increases the selectivity.

(29) The insensitivity or dead zone may be combined with the squaring in a characteristic curve. In the same way, it is possible to use the absolute amount of the high pass output signal, and it is also possible to combine the absolute value generation with the dead zone in a characteristic curve. These variants are shown in FIGS. 4a, 4b as well as 5a and 5b.

(30) FIG. 4a shows a filter unit 32 having an insensitivity region (dead zone) without step change, in which the characteristic curve is configured as a parabola, so that multiplier 24 may be omitted since the parabola-shaped transfer function of filter unit 32 already generates a squared signal on the output side.

(31) FIG. 4b shows a filter unit 32 having an insensitivity region (dead zone) without a step change, in which the absolute-value generation is combined with the dead zone in a characteristic curve.

(32) FIGS. 5a and 5b show systems as alternatives to those of FIGS. 4a and 4b, in which the characteristic curve of filter unit 33 features a step change in the insensitivity region.

(33) FIG. 6 shows a corresponding system, in which, in a deviation from the system shown in FIG. 2, integrator 25 divides the result by integration period T in each case, which means that the average signal powers 25.2 for P.sub.S and P.sub.M are available for the comparison. The comparison then takes place as follows:
.sub.M<.sub.unten und .sub.S>.sub.oben.fwdarw.sensor is oscillating

(34) Such a power comparison is of course combinable with all previously mentioned insensitivity regions according to FIGS. 3a, 3b, 4a, 4b as well as 5a and 5b.

(35) The use is conceivable both in gasoline or diesel combustion engines that require an oscillation detection of gas sensors, as may be the case in particular in sensors that are relevant for the exhaust gas. This monitoring function may be used as an autonomous function or combined with dynamic diagnosis functions, e.g., as described in German patent document 10 2011 088 296.0, filed on Dec. 12, 2011.