METHOD AND DEVICE FOR EVALUATING MEASURED VALUES DETERMINED DURING THE PRACTICAL DRIVING OF A VEHICLE

Abstract

The object of the present invention is to further improve the evaluation of measured values ​​obtained during the practical operation of a vehicle. This object is achieved by determining the time delay between a signal change at the input of the system, which comprises an internal combustion engine with an exhaust pipe and a mobile exhaust gas analyzer, and a corresponding signal change at the output of this system, by applying two independent methods or approaches in parallel. On the one hand, according to a first approach, this time delay is determined by means of a cross-correlation analysis. On the other hand, according to a second approach, this time delay is determined based on system identification, i.e.,by determining the dependence of the system's output variables on the system's input variables, which includes the combustion engine, the exhaust system / exhaust pipe and the (mobile) exhaust gas analyzer used, in conjunction with a balancing calculation, i.e., parameter identification / estimation or regression analysis.

Description

[0001] The present invention relates to a method and a device for evaluating measured values ​​obtained during the practical driving operation of a vehicle, with the features of the patent claims.

[0002] According to document DE102014006319A1, it is state of the art to measure and record exhaust emissions from an internal combustion engine during real-world driving (Real Driving Emissions, RDE) and to assign them to a specific vehicle operating state, thus enabling the determination of the causes of the exhaust emissions. This involves considering an exhaust gas transit time and an exhaust gas analysis time. This ensures that measured values ​​concerning exhaust emissions are assigned to the vehicle state that existed at the time of measurement by an exhaust gas analysis device. In particular, the measured values ​​concerning exhaust emissions are correlated with signals from the control and / or regulation of the internal combustion engine. The exhaust gas transit time is determined either based on generally known physical relationships, i.e.,The dimensions of the exhaust system and the prevailing volume flow are calculated or determined by adding an additive to the exhaust gas in conjunction with a time measurement.

[0003] The object of the present invention is to further improve the evaluation of measured values ​​obtained during the practical driving operation of a vehicle.

[0004] This problem is solved according to the invention by means of a method and a device according to the claims. According to the invention, the determination of the exhaust gas transit time and, if applicable, the exhaust gas analysis time or the time delay between a signal change at the input of the system, which comprises an internal combustion engine with an exhaust pipe and a mobile exhaust gas analyzer, and a corresponding signal change at the output of this system, is carried out by a parallel application of two independent methods or approaches.

[0005] On the one hand, according to a first approach, this time delay is determined using a cross-correlation analysis. On the other hand, according to a second approach, this time delay is determined based on or in conjunction with system identification, in particular by combining a theoretical and an experimental determination of the dependence of the system's output variables on the system's input variables. This system comprises the combustion engine, the exhaust system / exhaust pipe, and the (mobile) exhaust gas analyzer used, in conjunction with a least-squares adjustment, i.e., parameter identification / estimation or regression analysis.

[0006] Due to the linearity assumption, the approximately proportional relationship between the injection quantity and the CO2 concentration in an internal combustion engine, particularly a diesel engine, is used to determine the runtime of both concepts. With the exception of particulate filter regeneration, exhaust aftertreatment has virtually no effect on CO2. Knowing the CO2 runtime, the runtime for the other emission types can be corrected using the known time offset from the other analyzers of the measuring device.

[0007] The key aspect is the combination of the two methods / approaches for determining the desired time shift, as well as the fact that the correlation method is implemented virtually in real time within an evaluation window during an emissions test in real-world driving conditions. It is not necessary to have access to all the measurement data; instead, the calculation is performed in real time (subsequently). Furthermore, the method is applied to the entire engine load and speed range to account for the various influencing factors of the emission peaks, depending on the engine's control parameters and the nonlinear (emissions) optimization strategy of the engine control unit. The correlation parameters determined in this way are then visualized for the user using appropriate visualization techniques.

[0008] It is advantageous that two redundant paths underlie the determination of the desired time shift. Furthermore, it is advantageous that the method according to the invention is carried out using real data from the RDE test drive. A prior reference measurement to determine the transit times with stimulating input signals or added additives, which would exceed the justified effort of the overall concept, is therefore unnecessary.

[0009] By using both signal and system theory approaches to determine independent transit times / time shifts, the probability of a correct transit time determination increases. In other words, validating the results of both methods enhances the robustness of the evaluation in exhaust gas tests of the type mentioned.

[0010] Based on the highly accurate, real-time exhaust gas transit times determined according to the two independent approaches, possibly including exhaust gas analysis times on the underlying system, it is possible to identify the potential causes of exhaust gas peaks and statistically determine the probability of each cause. The high degree of non-linearity in the cause-and-effect relationship is mapped and visualized on the engine map (in real time). This enables a real-time analysis of emission peaks, which is of interest both for use with measurement technology for parameterizing the engine control unit and could theoretically also be implemented as a real-time function in the engine control unit to reduce emission peaks as an adaptation function.

[0011] Furthermore, a device is provided which is set up to carry out the method according to the invention.

[0012] Further advantageous embodiments of the present invention can be found in the following exemplary embodiment and in the dependent patent claims.

[0013] Particularly in connection with type approval, a vehicle's real-world emissions (Real Driving Emissions, RDE) are tested on the road. This type of on-road emissions test uses a portable emissions measurement system (PEMS) to simulate real-world driving conditions. The measured values ​​recorded and stored during this practical / actual driving test (in time steps) primarily relate to the vehicle's exhaust emissions, specifically those of the combustion engine powering the vehicle. As is well known, these emissions include, in particular, the concentrations of CO, CO2, HC, NO, NO2, and particulate matter present in the exhaust gas. See, for example, the generally known legal requirements, including the consolidated version of Regulation (EU) 2017 / 1151.

[0014] During such an emissions test, input and / or output signals from one or more of the vehicle's control units, in particular the engine control unit of the internal combustion engine, are continuously determined and recorded (i.e., measured and stored) at intervals; see also, for example, the consolidated version of Regulation (EU) 2017 / 1151 ("Engine control unit signals and data"). A possible output signal from the engine control unit relates to / represents the injection quantity. Input signals from the engine control unit include, among others, engine speed and load.

[0015] The measured values / data concerning the exhaust emissions of a vehicle and the measured values / data concerning (input and / or output) signals of a (engine) control unit of the vehicle (and / or the combustion engine) are recorded, in particular, synchronously in time (each in time steps, at discrete points in time, in a specific measurement grid, for example every 10 milliseconds).Figure 1 The image shows an excerpt from such a time-synchronized determination / recording of the injection quantity ("signals and data from the engine control unit") and the CO2 concentration present in the exhaust gas (of the vehicle / internal combustion engine). These measurement points (connected by a curve) were determined during a reference measurement at a crankshaft speed of 3500 revolutions per minute, during which three successive load changes from 0% to 30% were performed. This resulted in three successive significant changes in the injection quantity and consequently three successive significant changes in the CO2 concentration present in the exhaust gas. The injection quantity and the CO2 concentration in the exhaust gas, determined by the mobile exhaust gas analyzer, thus serve as reference signals and can be determined, for example, even before an exhaust gas test on a (stationary) test bench.

[0016] There is a strong relationship between these two quantities; as is known, there is a causal relationship between them, meaning there is a (true) correlation. Therefore, this relationship is predictable. However, other relationships or cause-and-effect relationships between measured values / data concerning a vehicle's exhaust emissions and measured values / data concerning (input and / or output) signals of the vehicle's (engine) control unit (and / or the combustion engine), which were determined and recorded during practical / actual driving, are not immediately clear or recognizable. For effective and targeted development, it is highly advantageous to make these relationships visible.In this way, it can be determined whether, and if so, what cause-and-effect relationships exist between measured values / data concerning a vehicle's exhaust emissions and measured values / data concerning (input and / or output) signals of the vehicle's (and / or the internal combustion engine's) (engine) control unit. In other words, it is extremely useful to be able to determine, virtually in real time, based on the measured values / data obtained and recorded during practical / actual vehicle operation, which parameters / data (input and / or output signals) of the respective (engine) control unit are responsible for various "emission peaks." Put another way, the targeted identification of relevant signal relationships is desirable.

[0017] However, as in Figure 1As shown, the determined injection quantity and the (time-synchronously / simultaneously) determined concentration of CO2 in the exhaust gas are (time-)shifted relative to each other, which of course also applies to all other measured values / measurement data concerning the exhaust emissions of a vehicle and concerning (input and / or output) signals of a (engine) control unit of the vehicle (and / or the) combustion engine.

[0018] This shift between the first reference signal, which can in principle be any input and / or output signal of a vehicle control unit (here, the "injection quantity"), and the second reference signal (specific concentration of an exhaust gas component, here CO2), results, among other things, from the time required for combustion and the exhaust gas transit time, as well as from the exhaust gas analysis time. The exhaust gas analysis time, or response time, i.e., the transfer characteristic or delay / time constant of the mobile exhaust gas analyzer used (T63 / T90 value), is generally known. However, the time required for combustion and the exhaust gas transit time from the engine (from at least one exhaust port or combustion chamber of the internal combustion engine) to the (respective) exhaust gas analyzer are not known.

[0019] The aim is therefore to determine the shift / time delay between the first reference signal and the second reference signal, i.e., the time interval between a signal change (output signal of the engine control unit - "injection quantity") at the system input (internal combustion engine) and the corresponding signal response (concentration of the component CO2 in the exhaust gas) at the system output (output of the exhaust gas analyzer).

[0020] According to the invention, this displacement is first determined using two independent methods or approaches.

[0021] One approach involves determining this shift using a cross-correlation analysis (cross-correlation function).

[0022] On the other hand, according to a second approach, this shift is determined on the basis of or in conjunction with system identification, in particular by means of a combination of a theoretical and an experimental determination of the dependence of the system's output variables on the system's input variables, which includes the combustion engine, the exhaust system / exhaust pipe and the (mobile) exhaust gas analyzer used, in conjunction with a adjustment calculation, i.e., parameter identification / estimation or regression analysis.

[0023] Regarding the first approach, the determination of a propagation delay / time offset, i.e., in this case, the determination of the shift described above, is carried out using cross-correlation analysis, specifically by calculating the product between two different signals and integrating it. As is known, this allows the measurement of signal coverage. The integral is calculated specifically over the variable shift of the output signal relative to the input signal. Thus, the correlation integral depends on the shift. The desired signal shift is found at the shift for which the correlation integral is maximal.

[0024] In the present case, the determination of the relevant propagation time / time offset of interest, i.e., the determination of the shift described above or the time interval between a signal change (output signal of the engine control unit – "injection quantity"; first reference signal) at the system input (internal combustion engine) and the corresponding signal response (concentration of the exhaust gas component CO2; second reference signal) at the system output (output of the exhaust gas analyzer) of the underlying system, is based on real measurement data acquired during the practical operation of a vehicle (in time steps, time-synchronized to each other). This information, i.e., the described shift, is preferably derived from the cross-correlation function, which includes the convolution of the impulse response and the autocorrelation function of the input signal, as will be described in detail below.The system is not stimulated with specific input signals, in particular with white noise or an approximation of such signals.

[0025] That is, in general, the cross-correlation function ϕ uy [I] φ uy τ = ∫ − ∞ ∞ y t u t − τ Δ t , where u the input signal (first reference signal; "injection quantity"), y the output signal (second reference signal, "concentration of the exhaust component CO2"), τ The time shift (the "shift" sought here, runtime / time offset) and t represent time (in general, in relation to the recording of the aforementioned measurement data).

[0026] For example, according to Föllinger, Otto: Regelungstechnik, Einführung in die Methoden und ihre Bewerbung, 11th edition; VDE-Verlag Berlin, 2013 .It is known that the cross-correlation function of the input signal and the output signal is equal to the convolution product of the impulse response of the transfer system or the underlying system and the autocorrelation function of the input signal. That is, if the system response of a linear time-invariant system [II] y t = g τ ∗ u t − τ = ∫ − ∞ ∞ g τ u t − τ Δ t where g represents the impulse response, then substituting [II] into [I] yields the relationship [III] φ uy τ = ∫ − ∞ ∞ g τ φ uu t − τ Δ t , i.e. the cross-correlation function ϕ uy ( τ ) results from the convolution of the impulse response and the autocorrelation function, and the desired shift lies at the maximum of ϕ uy ( τ ).

[0027] With regard to these generally known connections, we would like to expressly refer to the many literature sources available to the person skilled in the art on the filing date concerning the derivation and application of cross-correlation analysis, with regard to the reproducibility of the present description.

[0028] Specifically, according to the invention, in accordance with the first approach presented here, the measured values / data relating to the exhaust gases or exhaust emissions of a vehicle, i.e. the output signals, are used. y of the transmission system with at least one internal combustion engine and an exhaust gas analyzer, in particular those in Figure 1 The concentration of CO2 present in the exhaust gas (second reference signal) and the measured values / data relating to the (input and / or output) signals of a (engine) control unit of the vehicle, i.e. the input signals u of the underlying system, in particular those in Figure 1The injection quantity shown (first reference signal) is used to determine the desired displacement based on equations [I] to [III]. τ or to determine the maximum of ϕ uy ( τ ) used.

[0029] Based solely on the now-known shift between the first reference signal ("injection quantity") and the second reference signal (concentration of the exhaust component CO2), it is possible to determine, virtually in real time, which parameters / data (input and / or output signals) of the respective (engine) control unit are responsible for different "emission peaks," using the measured values / data obtained and recorded during practical / actual vehicle operation. This means that the now-known shift between the first and second reference signals can be applied to any measured values / data obtained and recorded during practical / actual vehicle operation concerning parameters / data (input and / or output signals) of the respective (engine) control unit, as well as concerning the vehicle's exhaust emissions measured (and recorded) using the mobile exhaust gas analyzer.applied so that the cause-and-effect relationships between measured values / data concerning the exhaust emissions of a vehicle and measured values / data concerning (input and / or output) signals of a (engine) control unit of the vehicle (and / or the) combustion engine can be recognized here as well.

[0030] For example, the recorded (control) signal concerning the exhaust gas recirculation rate (output signal of the engine control unit) can be shifted using / in analogy to the now known shift between the first reference signal ("injection quantity") and the second reference signal (concentration of the exhaust gas component CO2) relative to the recorded signal of the NOx concentrations (NO+NO2) present in the exhaust gas, so that, for example, it can be recognized that there is a causal signal relationship here, i.e., a change in the exhaust gas recirculation rate causes an undesirable strong increase in the NOx concentrations (NO+NO2) present in the exhaust gas, which must be prevented by a calibration of the engine control functions in the further course of events.

[0031] In one implementation of determining the desired displacement using a cross-correlation analysis (cross-correlation function), the desired displacement / runtime is determined by applying equations [I] to [III] prior to the cross-correlation analysis. τ or to determine the maximum of ϕ uy ( τ ) filtering or (temporal) manipulation of the input signal uor the first reference signal with the response / transmission behavior of the mobile exhaust gas analyzer used, where the exhaust gas analyzer exhibits, in particular, the response behavior of a PT1 transfer element. The desired shift is then determined by means of a cross-correlation analysis using the filtered first reference signal and the (unaffected) second reference signal. That is, the desired shift between the system's input signals (e.g., injection quantity) and output variables (concentration of specific exhaust gas components at the exhaust gas analyzer output) is determined based on a pre-conditioned (input) signal.As a result, the cross-correlation analysis then only determines the transit time caused by the need to transport the exhaust gas from the internal combustion engine (including the very short and therefore negligible time required for combustion) to the exhaust gas analyzer, i.e. the exhaust gas transit time from the internal combustion engine to the (inlet of the) exhaust gas analyzer(s).

[0032] In a subsequent determination of which parameters / data (input and / or output signals) of the respective (engine) control unit are responsible for different "emission peaks", it is of course necessary to consider both the exhaust gas transit time from the internal combustion engine to the (input of the) exhaust gas analyzer(s) as previously described, and the parameters used for filtering or (temporal) influencing the input signal. uor the response behavior / transmission behavior of the mobile exhaust gas measuring device / the respective exhaust gas analyzer used, i.e., the response time of the exhaust gas analyzer, must be taken into account.

[0033] In other words, exhaust gas runtime and exhaust gas analysis time must be considered (added), or the recorded measured values / data relating to quantities / data (input and / or output signals) of the respective (engine) control unit must be filtered with the response / transmission behavior of the mobile exhaust gas measuring device / exhaust gas analyzer used, and furthermore, the exhaust gas runtime from the internal combustion engine to the (input of the) exhaust gas analyzer(s), determined as described above, must be taken into account for a targeted identification of relevant signal relationships between the input and / or output signals relating to the respective control unit and the measured values / data relating to the exhaust emissions of a vehicle.

[0034] However, in connection with the determination of the shift between the first and second reference signals described above, a smoothing effect occurs in the impulse response. gon, which the conciseness of the correlation maximum (of ϕ uy ( τ )) reduced and thus contributes to increasing inaccuracy in determining the displacement. Due to the inaccuracy of the first approach in conjunction with a cross-correlation analysis, the invention further provides for a determination of this displacement also based on a further approach, namely a system identification, so that the result is two values ​​of the displacement sought.

[0035] This means that the findings obtained using the first approach are verified against those obtained using a second approach. With regard to system identification, this specifically involves identifying a suitable / correct transfer function for the assumed / underlying system, comprising an internal combustion engine, an exhaust system / pipe, and a (mobile) exhaust gas analyzer.

[0036] Taking into account the in Figure 2 The three transfer elements shown (P element relating to combustion, dead-time element relating to the exhaust system, PT1 element relating to the exhaust gas analyzer) result in the following transfer function in the image area. G ( s ) for the system with U(s) as an input signal (first reference signal; "injection quantity") and with Y ( s ) as output signal (second reference signal; "concentration of the exhaust gas component CO2"): G s = Y s U s = K ∗ e − sTt ∗ 1 / 1 + T 63 ∗ s , where KThis corresponds to a proportionality factor, since for combustion in the combustion engine a linear system behavior is assumed between the output signal of the engine control unit and the quality of the exhaust gas or the concentrations of individual components, especially CO2 in the exhaust gas, and the exhaust system is considered from a systems engineering point of view as a pure dead-time system with regard to (raw) emissions, and it is assumed that the exhaust aftertreatment systems have no relevant influence on the emission signal.

[0037] The transfer function contains the (yet to be determined) parameters of the proportionality factor. K , the dead time Tt and the (possibly known in advance) analyzer time constant T63 In summary, this transfer function, after inverse transformation into the time domain, corresponds to the following differential equation: Y t = K ∗ U t − Tt − T 63 Y ˙ t , where U the input signal (first reference signal; "injection quantity"), Ythe output signal (second reference signal, "concentration of the exhaust component CO2").

[0038] The discrete difference equation needed to identify the (system) parameters is given by the index i , the sampling time Δ t and an additionally introduced offset for the emission concentration of the ambient air yo : y i Δ t = yo + Ku i − Tt Δ t − T 63 y i Δ t − y i − 1 Δ t Δ t .

[0039] According to Figure 3 The relationships are shown again, but only schematically. Figure 1 The first and second reference signals are shown to illustrate the subsequent steps in determining the desired displacement using the second approach. In the upper diagram, the first reference signal is the injection quantity, i.e., the input signal. u of the system. In the lower diagram, the second reference signal is the concentration of the exhaust component CO2, i.e., the output signal. y shown. The in Figure 3 The signals shown are as in the context of Figure 1 explained and provided.

[0040] The determination of the desired shift using the second approach is characterized by the fact that parameters of the difference equation (3) are estimated by means of a least squares adjustment, i.e. a regression analysis (linear regression, least squares method), starting from the measured / recorded data of the first reference signal. u and the second reference signal y , including the (initially still present, see explanation below) system-related shift (exhaust gas runtime plus analysis time) between the first reference signal u and the second reference signal y .

[0041] This is achieved through an iterative, i.e., step-by-step shift, meaning a variation of the position of the first reference signal. u and the second reference signal yto each other, in particular the first reference signal u compared to the second reference signal y is shifted, whereby for each iteration (s-step) a (linear) parameter identification of the difference equation (3) or identification of parameters of the difference equation (3) is carried out.

[0042] Varying the position of the first reference signal u and the second reference signal yThe relationship between the data points is determined in particular by the time steps or discrete time points / measurement grid according to which the measured values / data concerning the exhaust emissions of a vehicle and the measured values / data concerning the (input and / or output) signals of a (engine) control unit of the vehicle (and / or the combustion engine) were each recorded (time-synchronously). That is, the first iteration step and the subsequent iteration step are offset from each other by one time step, for example, by 10 milliseconds.

[0043] Because the first reference signal u and the second reference signal y (Initially / at first) they are shifted relatively far apart, namely in the first step by the still unknown system-related shift, as in Figure 1 As shown, the adjustment of the second reference signal is successful. y the model according to the difference equation (3) only provides a small or insufficient result.

[0044] Consequently, the estimation of parameters of the difference equation (3) or the applied adjustment / regression analysis causes relatively large deviations. That is, the errors, error terms, or residuals, or measures based on or derived from the residuals, are comparatively large (in absolute value).

[0045] In other words, the estimation of parameters of the difference equation (3) is still so inadequate that the deviation or difference between the second reference signal modeled on the basis of the difference equation (3) y (Estimated value) which arises when approximate solutions of the parameters of the difference equation (3) are substituted into the difference equation (3) (in conjunction with the shifted first reference signal) u ), and the (according to Figure 1 ) measured second reference signal (measured value), i.e. the residual, is comparatively large.

[0046] That is, the magnitude of the deviation is relatively large. In the context of the present invention, this deviation corresponds, for example, to the (average or maximum) error, error term, or residual, or to a measure derived therefrom, such as, in particular, the residual sum of squares / coefficient of determination or the (especially squared) sum of the error terms. As is generally known in statistics, the residual sum of squares denotes the sum of the squared (least squares) residuals, i.e., deviations between observed values ​​(here, the measured second reference signal). y according to Figure 1 ) and the predicted values ​​(here, the second estimated reference signal) y ) all observations.

[0047] That is, within the framework of a model validation or analysis of the (remaining or resulting) error terms / residuals carried out during or after the respective iteration step, it becomes apparent that the deviations / residuals are initially still large, in the first iteration step of a variation of the position of the first reference signal. u and the second reference signal y relative to each other, in the respective parameter identification of the difference equation (3), starting from the initial state of a displacement / position of these two signals relative to each other, i.e. starting from the identification measurement according to the Figure 1 and 3 .

[0048] Since, in the further course of events, i.e., in the next iteration step, as in Figure 3 indicated by a horizontally oriented arrow or dashed lines, as a result of a further variation in the position or displacement of the first reference signal. uand the second reference signal y to each other, the correspondence between the first reference signal u and the second reference signal y By increasing the value, the second reference signal is successfully adjusted in this (second or further) iteration step and also in the subsequent iteration steps. y The model according to the difference equation (3) gets better and better, so that the estimation of parameters of the difference equation (3) produces smaller deviations compared to the previous iteration step, i.e. the errors, error terms or residuals or measures based on the residuals become comparatively smaller.

[0049] In the subsequent iteration steps, at one iteration step, in particular the iteration step where the greatest possible agreement between the first reference signal is achieved, a certain result is obtained. u and the second reference signal yexists, a minimum with respect to the errors, error terms or residuals or measures based on the residuals with respect to the adjustment calculation / regression, since the fitting of the second reference signal y the model according to the difference equation (3) is best achieved in this iteration step compared to the further iteration steps.

[0050] In practice, therefore, only the parameters are considered in the applied adjustment calculation / regression analysis. K (proportionality factor) and T63 (analyzer time constant) of the difference equation (3) were estimated. Provided the analyzer time constant T63 If only the parameter is known in advance, only the parameter will be used. K The proportionality factor of the difference equation (3) was estimated. That is, the parameter relating to the dead time. Tt is extracted from the adjustment function / regression function or obtained using the described application of adjustment calculation / regression.

[0051] In practice, this results in the dead time. Tt or the desired (system-related) shift of the first reference signal u and the second reference signal results. y in relation to each other in the further course of the process based on the model validations carried out in the individual iteration steps or the respective analysis of the residuals (or derived comparison quantities) that result from the parameter estimates of the difference equation (3) in the individual (iteration) steps or the individual variations of the position of the first reference signal u and the second reference signal y result in each other.

[0052] That is, the dead time Tt or the sought-after (system-related) shift of the first reference signal u and the second reference signal y The relationship between them now results from the number of iteration steps, which start from the in Figure 1shown, as yet unknown, system-related shift of the first reference signal u and the second reference signal y until the iteration step was found in which such a or greatest possible agreement between the first reference signal was achieved. u and the second reference signal y consists of such that a minimum results with regard to the errors, error terms or residuals or measures based on the residuals, since the fitting of the second reference signal y The model according to the difference equation (3) is most successful in this iteration step compared to the subsequent iteration steps. The dead time is then determined. Tt or the sought-after (system-related) shift of the first reference signal u and the second reference signal yThe differences between them are not only due to the number of iteration steps described above, but also, of course, to the time interval between the individual iteration / variation steps, which are offset from each other by one time step, for example by 10 milliseconds.

[0053] In other words, the second reference signal is determined using the difference equation (3). y or its formation rule described, specifically depending on the first reference signal. u as well as depending on the still unknown and yet to be determined shift between the first reference signal u and the second reference signal y , whereby the shift results from the still unknown dead time Tt and the (possibly previously known) analyzer time constant T63.

[0054] In conjunction with a regression analysis or adjustment calculation, the system-related shift resulting from an identification measurement between the first reference signal is then calculated. u and the second reference signal y according to the Figure 1 and 3 , in the further course of events, in conjunction with a stepwise variation of the shift between the first reference signal u and the second reference signal y , i.e., a change in the position of the first reference signal that is, in a sense, a test or interim change that occurs in individual steps. u compared to the second reference signal y and the individual steps then present (until the best possible coverage / match of the first reference signal is achieved). u and the second reference signal y (reducing and then increasing) shift between the first reference signalu and the second reference signal y , a model of the relationship between the (dependent) second reference signal y and the first reference signal (corresponding to the independent variable) u based on the difference equation (3) by determining the parameters of the difference equation (3) and an evaluation or verification of whether the parameters of the difference equation (3) obtained by means of the adjustment calculation / regression analysis in conjunction with the first reference signal u cause / produce a deviation (an error / a residual or a derived quantity) that differs from the deviation according to the previous or subsequent (iteration) steps with a different position of the first reference signal. u compared to the second reference signal y, smaller or minimal, whereby, if this minimum is known, it is possible to determine or locate the size of the desired, system-related shift between the first reference signal u and the second reference signal. y is.

[0055] In other words, the second approach to determining the parameters of the difference equation (3) uses linear regression. The parameter of dead time Tt Difference equation (3) represents a nonlinear system. To still be able to use linear regression, the dead time parameter must be Tt from the regression function. This means that the estimation of the system parameters is only possible for the remaining, linear parameters ( K, T63 ) is performed. This linear system identification is iteratively performed for all output signal shifts to be investigated. Y(second reference signal) performed. The shift at which the linear regression exhibits its minimum residual sum corresponds to the desired dead time. Tt of the system. Additionally, the time constant can be used in the regression method. T63 can be specified when the response behavior of the analyzers is known. Thus, prior knowledge of the system, such as that available for exhaust gas analyzers, can be used for parameter identification. At the beginning of the runtime determination, the signal dynamics of the output signal(s) are evaluated to exclude inaccurate results due to insufficient signal dynamics. The dynamics evaluation reflects the relative dispersion of the signal(s) under consideration within the evaluation window. If this exceeds the predefined threshold, further evaluation is triggered. In a loop, the shift of the signals within the runtime interval under investigation is incremented. For each iteration, the linear parameter identification of the difference equation is performed. In the event that the time constant of the PEMS analyzer ( T63If the time constant is known, it is specified in the parameter estimation. Otherwise, it can also be determined using the system identification. The desired shift of the output signal, i.e., the shift of the second reference signal y relative to the first reference signal, is then calculated. u The minimum residual sum is reached at the time of the complete traversal of the shift interval. After the entire shift interval has been traversed, the propagation delay is determined by cross-correlation of the (preferably) filtered input signal with the output signal, according to the first approach. The cross-correlation function is evaluated for local maxima within the interval of propagation delays under investigation.

[0056] As a result, the findings of the determination of the desired displacement using two independent methods or approaches are thus available for further processing. That is, according to the invention, after the dual / parallel determination of the desired displacement, a comparison of the results—i.e., the displacements determined using the two independent methods / approaches—is performed. The displacement determined using the local maxima of the cross-correlation function between signals at the input of the underlying system (comprising the internal combustion engine, exhaust system, and exhaust gas analyzer) and signals at the output of this system is compared with the displacement provided by the system identification.If a predefined tolerance is maintained between the signal shifts provided by the two independent approaches, the determination of the previously unknown shift is considered successful. Subsequently, either only one of the two determined shifts can be used as the basis for further processing, or both independently determined shifts can be processed, for example, a value derived from the two independently determined shifts, such as an average of the two independently determined shifts.

[0057] In other words, these two available values, determined on different bases, regarding the desired shift / time delay between the first reference signal ("injection quantity") and the second reference signal (concentration of the exhaust gas component CO2) are subsequently compared. Based on this comparison, a decision is made as to whether the deviation between the two shifts lies within a predefined tolerance. That is, if a tolerance for the shifts / dead times between the two methods is maintained, or if the shifts / dead times are sufficiently similar, these two determinations are considered successful, and one or both of the determined shifts can be further processed.The analysis can be based on other signal relationships, for example, the causal signal relationship described above between the signal concerning the exhaust gas recirculation rate (output signal of the engine control unit) and the signal of the NOx concentrations present in the exhaust gas (NO+NO2). Thus, according to the invention, the probability of a correct determination of the shifts / dead times is increased (through the validation described above), and the robustness of subsequent work concerning the calibration of the control and / or regulation of the underlying system is improved. If this tolerance is not maintained, the results of the shift determination according to the two approaches are discarded, and, for example, the shift is determined again according to the two approaches.

[0058] In any case, based on the shift determined using the second approach and thus known, between the first reference signal ("injection quantity") and the second reference signal (concentration of the exhaust component CO2), it is also possible (as described in connection with the first approach) to determine virtually in real time which parameters / data (input and / or output signals) of the respective (engine) control unit are responsible for different "emission peaks". This means...This shift between the first reference signal and the second reference signal, determined by other means and now known, is also transferred or applied to any measured values / data obtained and recorded during the practical / actual operation of a vehicle concerning quantities / data (input and / or output signals) of the respective (engine) control unit, as well as concerning the exhaust emissions of the vehicle determined (and recorded) by means of the mobile exhaust gas measuring device, so that here too the cause-and-effect relationships between measured values / data concerning the exhaust emissions of a vehicle and measured values / data concerning (input and / or output) signals of a (engine) control unit of the vehicle (and / or the) combustion engine can be recognized.

[0059] The device proposed according to the invention, which is configured to carry out the method according to the invention, comprises in particular a computer equipped for carrying out the method according to the invention, with a CPU and a machine-readable storage medium, wherein a computer program is stored on the storage medium which includes all features or steps of the method according to the invention, and wherein the computer program is executed by means of the CPU. This computer program or software tool for calibration / application work is connected, in particular via CAN, to the hardware of the PEMS and to a control unit of the vehicle using suitable interface modules.In any case, the results of the verification of the findings obtained using the first approach and the findings obtained using the second approach are displayed to an operator via a graphical user interface, in particular whether the determination of the previously unknown displacement was successful.

Claims

1. A method for evaluating measured values ​​obtained during the practical operation of a vehicle, wherein: - the time delay between a signal change at the input of a system comprising an internal combustion engine with an exhaust pipe and a mobile exhaust gas analyzer, and a corresponding signal change at the output of this system is determined by the parallel application of two independent methods; - the two resulting time delay values ​​are compared, and a decision is made as to whether a deviation between the two time delays is within a specified tolerance or not, depending on this comparison; - if the deviation is within the tolerance, the determination of the time delay is considered successful, and at least one of the two time delays is further processed.so that, with knowledge of this time delay, measured values ​​obtained by means of the exhaust gas analyzer can be assigned to those signals of the control and / or regulation of the combustion engine which were present at the time the measured values ​​were obtained by means of the exhaust gas analyzer; if the deviation is not within the tolerance, the determination of the time delay is discarded and a new determination of the time delay is carried out by a parallel application of the two independent methods.

2. Method according to claim 1, wherein according to the first method the time delay is determined by means of a cross-correlation analysis and according to the second method the time delay is determined on the basis of a system identification by determining the dependence of the output variables of the system on the input variables of the system.

3. Method according to claim 2, wherein the determination of the time delay according to the first method and according to the second method is carried out using a first reference signal and a second reference signal.

4. Method according to claim 3, wherein the first reference signal is an output signal of the control and / or regulation of the internal combustion engine, which describes the proportion of fuel supplied to the internal combustion engine and which is present at the system input, and the second reference signal is an output signal of the exhaust gas analyzer, which is the signal response at the system output corresponding to the first reference signal and which describes the exhaust gas component carbon dioxide (CO2) in the exhaust gas of the internal combustion engine.

5. Device configured to carry out the method according to claims 1 to 4.

6. Device according to claim 5, characterized by the fact thatA computer equipped for carrying out the method according to one of claims 1 to 4 is provided with a CPU and a machine-readable storage medium, wherein a computer program is stored on the storage medium which comprises all steps of a method according to one of claims 1-4, wherein the computer program is executed by means of the CPU.

7. Computer program that performs all steps of a method according to any one of claims 1-4 when run on a computer.

8. Computer program product comprising program code stored on a machine-readable storage medium for carrying out the method according to any one of claims 1 to 4 when the program is executed on a computer.

9. Vehicle with a device according to claims 5 to 8.