WO2021092639A1 - Procédé et système d'analyse et/ou d'optimisation d'une configuration d'un type de véhicule - Google Patents

Procédé et système d'analyse et/ou d'optimisation d'une configuration d'un type de véhicule Download PDF

Info

Publication number
WO2021092639A1
WO2021092639A1 PCT/AT2020/060399 AT2020060399W WO2021092639A1 WO 2021092639 A1 WO2021092639 A1 WO 2021092639A1 AT 2020060399 W AT2020060399 W AT 2020060399W WO 2021092639 A1 WO2021092639 A1 WO 2021092639A1
Authority
WO
WIPO (PCT)
Prior art keywords
configuration
base
vehicle
operating
data set
Prior art date
Application number
PCT/AT2020/060399
Other languages
German (de)
English (en)
Inventor
Kevin LAUBIS
Sascha Bauer
Original Assignee
Avl List Gmbh
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Avl List Gmbh filed Critical Avl List Gmbh
Priority to DE112020004941.8T priority Critical patent/DE112020004941A5/de
Publication of WO2021092639A1 publication Critical patent/WO2021092639A1/fr

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/10—Geometric CAD
    • G06F30/15—Vehicle, aircraft or watercraft design
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F02—COMBUSTION ENGINES; HOT-GAS OR COMBUSTION-PRODUCT ENGINE PLANTS
    • F02D—CONTROLLING COMBUSTION ENGINES
    • F02D41/00—Electrical control of supply of combustible mixture or its constituents
    • F02D41/24—Electrical control of supply of combustible mixture or its constituents characterised by the use of digital means
    • F02D41/2406—Electrical control of supply of combustible mixture or its constituents characterised by the use of digital means using essentially read only memories
    • F02D41/2425—Particular ways of programming the data
    • F02D41/2429—Methods of calibrating or learning
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M17/00—Testing of vehicles
    • G01M17/007—Wheeled or endless-tracked vehicles
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/10—Geometric CAD
    • G06F30/17—Mechanical parametric or variational design
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F02—COMBUSTION ENGINES; HOT-GAS OR COMBUSTION-PRODUCT ENGINE PLANTS
    • F02D—CONTROLLING COMBUSTION ENGINES
    • F02D2200/00—Input parameters for engine control
    • F02D2200/02—Input parameters for engine control the parameters being related to the engine
    • F02D2200/08—Exhaust gas treatment apparatus parameters
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/20—Pc systems
    • G05B2219/23—Pc programming
    • G05B2219/23005—Expert design system, uses modeling, simulation, to control design process
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/18—Manufacturability analysis or optimisation for manufacturability
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G06N20/20—Ensemble learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound

Definitions

  • the present invention relates to a method and system for analyzing and / or optimizing a configuration of a vehicle type, with a configuration to be tested of a vehicle of the vehicle type being operated in an RDE test mode.
  • the road as a test bench environment offers the necessary stochastic basis to ensure that motor vehicles also comply with the required emission targets when used by customers.
  • the influences that are difficult to control it is almost impossible to carry out two measurements with comparable conditions during real test drives on the road.
  • the effects of changes to a vehicle, in particular to a Drivetrain or an exhaust system do not specifically compare with a basic state or a basic configuration. This makes it difficult to make a statement about the effectiveness of changes.
  • the road has so far only been regarded as suitable to a limited extent as a development environment.
  • One approach is to generate dynamic speed profiles which enable the reproduction of a test operation on test benches or based on models, which at least essentially corresponds to operation in real road traffic.
  • the transformation model having an assignment rule between the observation variables and a property of at least one device, the transformation model being set up on the basis of at least one data record recorded on the unit to be tested in the defined operating cycle to output an expression of the at least one device to be analyzed.
  • a first aspect of the invention relates to a computer-aided method for analyzing and / or optimizing a configuration of a vehicle type on the basis of an RDE test operation, in particular for comparing a configuration to be tested with a basic configuration, comprising the following work steps:
  • a second aspect of the invention relates to a method for training a vehicle model for simulating an operating behavior of a vehicle, comprising the following work steps:
  • a third aspect of the invention relates to a system for analyzing and / or optimizing a configuration of a vehicle type on the basis of an RDE test operation, comprising:
  • Data processing means for recording a first data set of operating variables of a configuration to be tested of a vehicle of the vehicle type and of environmental variables as a function of a distance covered and / or a past time continuously during an RDE test operation of the configuration to be tested, part of the first data set showing the operating behavior of the characterizes the configuration to be tested under certain operating conditions;
  • Simulation means for simulating an operating behavior of a basic configuration of a vehicle of the vehicle type by means of a vehicle model of the basic configuration, wherein when simulating, variables of the recorded first data set are input variables of the vehicle model and a second data set of at least one operating variable of the basic configuration is generated, and the second data set contributes to the operating behavior of the basic configuration characterizes the specific operating conditions; and comparison means for comparing the performance of the configuration under test with the performance of the basic configuration on the basis of the portion of the first data set and the second data set.
  • a compensation calculation in the sense of the invention is preferably a regression method or a pattern recognition method, in particular on the basis of a neural network or a random forest regression, and / or a combination of both, in particular in terms of statistical teaching.
  • a simulation within the meaning of the invention can be carried out on a test bench or purely based on models on a computer.
  • at least one component can preferably also be operated on a test stand in a simulated operation and at least one other component can be operated on a computer based on a model.
  • a simulation of an operating behavior within the meaning of the invention is preferably a time-resolved or distance-resolved simulation of the operating behavior of a vehicle.
  • a vehicle's operating behavior is determined in particular by operating parameters such as speed, torque, mass flows, pressures, temperatures, in particular coolant temperature, fuel consumption, consumption of other operating resources, emissions, OBD values, speed, gear selection, emissions, in particular C0 2 emissions, nitrogen emissions , Particle number emission etc., characterized.
  • Operating conditions within the meaning of the invention are preferably composed at least of driving style and environmental conditions.
  • Ambient variables within the meaning of the invention are preferably road gradient, height above zero level, absolute humidity, ambient pressure, ambient temperature.
  • output preferably means providing data. In particular, this can be done at a data interface and / or at a user interface.
  • a means within the meaning of the invention can be designed in terms of hardware and / or software and in particular a processing unit, in particular a microprocessor unit (CPU) and / or a data or signal connected to a memory or bus system, in particular a digital processing unit, in particular a microprocessor unit (CPU) and / or a have several programs or program modules.
  • the CPU can be designed to process commands that are implemented as a program stored in a memory system, to acquire input signals from a data bus and / or to give output signals to a data bus.
  • a storage system can have one or more, in particular different, storage media, in particular special different storage media, in particular optical, magnetic, solid and / or other non-volatile media.
  • the program can be designed in such a way that it embodies or is capable of executing the methods described here and that the CPU executes the steps of such methods.
  • a vehicle type within the meaning of the invention is preferably a set of vehicles which match in their essential features and are therefore preferably structurally identical.
  • the essential components, in particular the drive, are preferably structurally identical in one type of vehicle.
  • a specific vehicle is therefore preferably a specific implementation of the vehicle category.
  • An operating behavior within the meaning of the invention is preferably characterized by values of the operating variables.
  • a configuration within the meaning of the invention preferably corresponds to a realization of the vehicle type which is characterized by an expression of at least one property of at least one of the devices of the vehicle.
  • a basic configuration within the meaning of the invention is preferably a defined configuration of the vehicle type from which a configuration to be tested was configured.
  • a driving style within the meaning of the invention is preferably a constellation of values of operating variables or a course of values of at least one operating variable which influence the operating behavior of a vehicle. Driving style is characterized in particular by the variables acceleration, throttle valve position and speed as a function of the course of the road and / or a driving situation.
  • An environment within the meaning of the invention is preferably an environment and / or a driving situation around a vehicle under consideration.
  • An RDE test operation within the meaning of the invention is preferably a test operation under RDE conditions.
  • an RDE test operation can also be any other type of test operation, in particular a random test operation.
  • An RDE test operation is preferably carried out under real conditions, in particular on the road.
  • a vehicle model within the meaning of the invention is preferably a model that depicts the operating behavior of a vehicle.
  • the vehicle model can preferably only calculate a single operating variable, which is a dependent variable in the sense of machine learning. Furthermore, such a vehicle model preferably has sub-models which can each calculate an individual operating variable.
  • the invention is based on the approach of making test operations, which were carried out with vehicles of a vehicle category with a different configuration, comparable by adjusting each size that is determined from outside the respective vehicle. These preferably characterize the driving style with which the vehicle is operated by a driver or also by a driver assistance system, and the surroundings of the vehicle.
  • the invention is based on the knowledge that an operating behavior of a configuration of a vehicle of the vehicle category to be tested can be traced back to an operating behavior of a vehicle of the vehicle category with the basic configuration, in that it is specified that the basic configuration under the same operating conditions as the test operation of the to configuration under test would have been performed.
  • a vehicle model is used here which is preferably based on a deductive method.
  • Machine learning algorithms such as, for example, artificial neural networks or random forest algorithms, are preferably used to form such a vehicle model.
  • One advantage of the invention is that the effect of a changed configuration (for example compared to the basic configuration) can be determined on the basis of fewer RDE test operations of a configuration to be tested or even a single test operation of the configuration to be tested. A statistically meaningful number of RDE test operations for the configuration to be tested is not necessary. Only the vehicle model of the basic configuration should be examined as completely as possible and thus be valid for a complete representation of the vehicle in the test room, which is set up by the parameters varied in the RDE test operations.
  • the invention can therefore significantly reduce the effort required for test drives with a configuration to be tested. Furthermore, evaluations of the operating behavior of the configuration to be tested in relation to different driving styles and environmental conditions can be objectified. Therefore, the test drives do not necessarily have to be carried out with different drivers.
  • the vehicle model has an assignment rule between operating conditions, in particular driving style and ambient conditions, on the one hand, and the operating behavior, on the other hand, the vehicle model being based on a compensation calculation, in particular an artificial neural network or a random Forest algorithm, based on a data set that results from several RDE test operations with the basic configuration, each with different operating conditions.
  • a separate compensation calculation for the vehicle model is preferably used for each operating parameter that characterizes the operating behavior.
  • the use of a compensation calculation to create the vehicle model makes it possible to take into account a very large number of operating variables and environmental variables.
  • the formation of physical models for a type of vehicle would be for the set the information to be processed is very complex and would at least require a very long period of time.
  • the second data set of operating variables of the basic configuration is also recorded as a function of a distance covered and / or a past period of time.
  • time segments of an RDE operation or even an entire RDE operation can be reconstructed for the basic configuration in order to determine the operating behavior of the basic configuration. These time segments can then be compared with the corresponding time segments of the operating behavior of the configuration to be tested.
  • a parameter is determined on the basis of the comparison and is output.
  • a parameter can contain an evaluation of the configuration to be tested in relation to the basic configuration.
  • the configuration to be tested and the basic configuration differ by aging of the catalyst or filling of a particle filter, the second data set containing at least values of an emission.
  • the operating behavior is essentially characterized by the emissions with different configurations of the vehicle's exhaust system.
  • the data area is selected using a feature selection method, preferably a filter-based feature selection method.
  • those features or characteristics for example operating parameters, are identified which are used in the model formation must be taken into account and which are also influenced by driving styles and environmental conditions when comparing the configuration to be tested and the basic configuration.
  • sliding influencing areas are formed in the data sets to identify features, which are to be examined in the following work steps for the correlation between environmental conditions, in particular driving style and environmental conditions, on the one hand, and operating behavior on the other. By considering entire areas of influence, the correlations for the operating behavior can be predicted more precisely.
  • the recorded operating variables and environmental variables are selected in such a way that they are independent of the respective configuration of the vehicle type.
  • This advantageous embodiment ensures that effects which are caused by a change in the configuration are not confused with effects which are caused by changed operating conditions.
  • the invention also relates to a computer program which comprises instructions which, when executed by a computer, cause the computer to carry out the steps of one of the methods according to the invention, and to a computer-readable medium on which such a computer program is stored.
  • 1 shows a combined representation of a method and system for analyzing and / or optimizing a configuration of a vehicle type
  • 2 shows a block diagram of a method for training a vehicle model to simulate an operating behavior of a vehicle
  • FIG. 3 shows an illustration of a procedure for identifying candidate sizes, also called feature extraction
  • FIG. 6 shows a multiple diagram with different sizes in relation to a configuration to be tes and a basic configuration.
  • test drives In order to examine the effects of configuration changes in vehicles, it is necessary to take into account different driving styles and environmental conditions during the execution of test operations, in particular test drives.
  • this represents a major challenge, especially in RDE test operations, since an increase in the number of interfering variables that cannot or at least not completely controllable, such as ambient humidity, temperature, traffic, shifting and acceleration behavior, etc., leads to an increase in the Amount of the number of required measurement data leads to the possible influences of the interfering variables in the measurement data.
  • This amount of data grows exponentially, the more confounding variables, i. H. Dimensions of an operating room.
  • the generation of such an amount of test data is limited by the amount of time required to perform test drives.
  • FIG. 1 shows a combined illustration of the method 100 for analyzing a configuration of a vehicle category and a system 10 which is set up to carry out such a method.
  • This method 100 uses the methodology of machine learning in order to enable a conclusion (inference) about an operating behavior of a configuration of a vehicle under certain operating conditions in which another configuration of the vehicle was tested.
  • the method 100 is based on a vehicle model 0 ase for a basic configuration 2 or an existing configuration of the vehicle type.
  • This vehicle model 0 base is determined by adapting model parameters in such a way that relationships between different measured operating variables and environmental variables are accurately reproduced by the vehicle model 0 base.
  • a number of test drives are carried out and the vehicle model 0 base is formed by means of a machine learning algorithm, in particular a random forest algorithm or an artificial neural network.
  • This modeling is described in detail with reference to FIG. 2.
  • Those variables which are taken into account in the modeling are preferably determined by means of a feature selection method.
  • Candidate sizes are extracted from the raw data of measurements (feature extraction). This is described in detail with reference to FIG. 3.
  • further boundary conditions are preferably taken into account, such as, for example, the dependency of the variables which are to be included as input variables in the vehicle model 0 base on the configurations themselves.
  • the vehicle model 0 base of the basic configuration 2 After the vehicle model 0 base of the basic configuration 2 has been determined, it can be used to analyze further configurations 1 of the vehicle type.
  • the vehicle model 0 base is used in particular to make the other configurations 1 comparable with the base configuration 2. This is because the test results X var of these further configurations 1 can be observed directly, but a meaningful comparison between these test results X var and the test results X base of the basic configuration 2 is not directly possible.
  • the method 100 solves this problem by ase means of the vehicle model O tester were results of (y base) traversed by a test or tests which tion with the further configura 1, simulating or be estimated as if these tests with the basic configuration 2 would have been executed.
  • RDE test operations are carried out with a single configuration of the vehicle type.
  • a basic configuration 2 of the vehicle category is used for this.
  • the various RDE test operations are carried out 201 with different test environments and possibly also different vehicles of the same basic configuration 2.
  • test results are recorded 202 as data record X base .
  • data record X base preferably contains measured operating parameters of basic configuration 2 and environmental parameters.
  • Operating variables are, for example, carbon monoxide emissions, carbon dioxide emissions, nitrogen emissions, particle emissions, acceleration, engine speed, engine torque, throttle valve position, speed, a cumulative distance traveled, lambda, coolant temperature, total exhaust gas pressure, exhaust gas temperature and catalyst temperature.
  • Ambient variables are, for example, road gradient, height above zero level, absolute humidity, ambient pressure, ambient temperature.
  • the operating parameters and environmental parameters are recorded as a function of a distance covered and / or a past period of time. Part y base of the The data record therefore characterizes the operating behavior of the at least one vehicle in the different test environments in which the test drives were carried out. The operating behavior is indicated in particular by the operating parameters.
  • a data set X ' base is selected from the data set X base , for which a correlation between operating conditions, in particular driving style and ambient conditions, on the one hand and the operating behavior on the other, can be determined.
  • Statistical methods are preferably used here.
  • the selection is preferably part of the data preprocessing for training the vehicle model 0 base .
  • selecting 203 at least one of the following methods is used: selecting independent channels; Feature extraction, feature selection.
  • the sequence is preferably as follows: the independent channels are selected, and a feature extraction is carried out on the basis of these channels. These extracted features are candidates for training the vehicle model 0 base , on the basis of which the feature selection is carried out. Details of the methods are explained as follows:
  • the independent channels ie the variables which are independent of a change in a configuration
  • the selection of the independent variables is preferably carried out on the basis of expert knowledge. This ensures that effects that are caused by a change in the configuration are not confused with effects that are caused by changed operating conditions. Depending on what changes are made to a basic configuration, the independent variables are generally different.
  • the data set X ' base (as described above) is characterized on the one hand by the fact that a correlation (ie a possible mutual dependency) between driving style and ambient conditions on the one hand and the operating behavior on the other hand can be determined; On the other hand, it is characterized by the fact that the data record X ' base itself is not different from the one to be examined Configuration change is dependent. Only those variables that are independent of a configuration change can later serve as input variables for the vehicle model. Otherwise, the configuration changes would not only have an impact on the operating behavior, but also on the operating conditions. In this case, however, it would not be possible to make a clear statement on the basis of the vehicle model 0 base as to how a basic configuration 2 would have behaved under other operating conditions.
  • FIG. 3 Corresponding dependencies are shown in FIG. 3. There must be no dependency between the configurations and the input variables of the vehicle model when simulating the operating behavior of a basic configuration. If these input variables were also influenced by a change in configuration, ie if the dependency in FIG. 3 were not crossed out, the effect of the configuration change would be reflected indirectly via the input variables in the model of the basic configuration when simulating. However, this should be avoided because the vehicle model 0 base should display the test result as it would be without a configuration change.
  • Feature extraction is described below with reference to FIG. 4. This feature extraction is preferably carried out only in relation to variables that are independent of a configuration change. In this case, feature extraction is only performed in terms of these quantities.
  • the feature extraction can also be carried out in relation to the entire data set X base or all sizes.
  • So-called features are extracted from the examined variables, in particular those that are independent of configuration changes, ie data areas in which a correlation between operating conditions on the one hand and the operating behavior on the other can be expected.
  • the operating conditions are preferably determined by the driving style and the ambient conditions during a test drive. Both the operating conditions and the operating behavior are each characterized by one or more variables, ie operating variables and / or environmental variables.
  • Fig. 4 in relation to a variable x and a farm variable y of a data set X, to find correlations between the variable x and the farm variable y sliding influence areas w are considered, with which preferably the potential influence of a whole sequence of Values of the size x is declared to a value of the farm size y.
  • One or more aggregate functions s are preferably used or formed for each influencing area w in order to extract scalar features which are assigned to a value of the operating variable y, which is intended to be dependent on the variable x. More than one size x is preferably taken into account in the extraction. Depending on the number of variables x and their properties, not only individual influencing areas w are taken into account, but a whole set of influencing areas W.
  • An aggregate function can, for example, be a minimum value, maximum value, mean value, etc. in the data area under consideration.
  • Feature extraction makes it possible to use information about a specific time period or a specific distance covered in the estimation.
  • the combination of time periods in areas of influence results in fewer features than would be the case if each individual value of the size x were considered individually. This reduces the number of variables x to be considered and thus the dimensionality, i.e. the number of dimensions, in the analysis.
  • an application with the name tsfresh ® can preferably be used. This enables features to be extracted on the basis of standard aggregate functions, such as statistics on a distribution, the frequencies contained or the number of specific events.
  • the number of features extracted can be large because it is proportional to the number of aggregate functions S and the number of areas of influence W.
  • an importance of features is preferably determined, ie their influence on the output size.
  • the importance of an extracted feature is preferably determined by checking the hypothesis that this feature is correlated with an operational variable y.
  • the so-called Pearson correlation can be used as the correlation statistic, which is a measure of a linear relationship between two variables and is determined via a parametric test.
  • Another correlation statistic is the Kendall's Tau. This can also preferably be used.
  • Another method that can be used here is the Benjamini-Hochberg method.
  • the vehicle model 0 base is trained 204 in a training phase by reading the selected data areas X ' base into a compensation calculation.
  • important features ie important features, are used as input variables in the vehicle model 6 base during training. These are transferred from the preprocessing phase (work step 203) to the training phase.
  • Examples of algorithms that can be used in the context of such a compensation calculation are linear regression, logistic regression, random forest algorithms or regressions, ensemble methods, support vector machines, algorithms that work on (deep, recurrent or convolution-based) artificial neural networks, in particular long-term memory algorithms, etc.
  • the use of so-called deep algorithms makes extraction of features superfluous. If a simulation is carried out using a vehicle model 6 base trained in this way and if input variables from a data record X base or an edited data record X ' var of the test drives with the basic configuration 2 are provided to this simulation, then the vehicle model 0 base reproduces that part y base des Dataset
  • X base approximately as a simulation result y base , which characterizes the operating behavior of configuration 2 in the test environment used for training. This is shown in FIG. 5.
  • the vehicle model 0 base gives the operating behavior for training the vehicle model 0 base , the basic configuration 2 used under those operating conditions that prevailed during the test drive of configuration 1 to be tested.
  • a test drive or several test drives with a vehicle of the modified configuration 1 of the vehicle type is / are first carried out.
  • the fact that the vehicle shown is the modified configuration is shown in FIG. 1 by the change symbol with two rectangles and two arrows.
  • the data set X var resulting from these measurements is analyzed by means of a method 100.
  • the first data set X var is recorded 101 with the measured operating parameters of the configuration 1 to be tested and the measured or provided environmental parameters in a data memory of data processing means 12 a past time duration of the test drive or test drives with the configuration 1 to be tested is stored.
  • a part y var of this first data set X var in particular a part of the variables, preferably characterizes an operating behavior of the configuration to be tested 1.
  • Another part of the first data set X var preferably characterizes the operating conditions which occur during the test drive (s) with the configuration to be tested 1 templates.
  • data preprocessing is preferably carried out which corresponds to a feature extraction.
  • the same aggregate functions are preferably used or formed over the same data areas as when generating or training the vehicle model 0 base .
  • the same data area is preferably that data area which corresponds to a time segment of the past period of time or a route segment of the traveled route in which a certain feature was extracted when generating or training the vehicle model 6 base.
  • This data preprocessing is also preferably carried out only for those extracted features which were selected in the feature selection when generating or training the vehicle model 6 base . This saves computing time.
  • a processed first data record X ' var emerges from the data preprocessing.
  • Starting var of the recorded first data set X or the processed first data X 'var is in a simulating means the performance of Ba siskonfiguration 2 of a vehicle of the type of vehicle simulates 103.
  • the simulation is hereby var only for those data areas of the recorded first data set X or processed first data set X ' var , in which base features were extracted when generating or training the vehicle model 6.
  • the simulation is preferably carried out exclusively for those features that were selected when generating or training the vehicle model 6 base . This can also save computing time.
  • one or more variables of the first data set X var are included in the simulation as input variables of the vehicle model 6 base or as input variables in this vehicle model 6 base .
  • the input variables of the vehicle model 6 base are selected in such a way that they have only a slight dependency or preferably no dependency at all on the configuration changes that exist between the configuration to be tested and the basic configuration.
  • the model 6 base uses the information from the data set X var from the real test drives, namely as independent features or operating conditions.
  • the vehicle model 0 base generates a second data set y base of at least one operating variable of the basic configuration 2. As shown in FIG. 1, the configuration 2 of the vehicle type is included in the simulation 103 via the vehicle model 6 base.
  • a virtual operating behavior of the basic configuration 2 can be mapped in the second data set y base under the operating conditions which are presented during the test drive or tests with the configuration 1 to be tested.
  • the basic configuration 2 preferably completes the same test drive or the same test drives as the configuration 1 to be tested, with which the first data record X var was generated.
  • An individual vehicle model 6 base preferably only depicts one operating variable with regard to the operating behavior, for example a CO 2 emission or a nitrogen oxide emission.
  • This company size is preferably a dependent variable in the sense of machine learning.
  • several models can also be trained; For example, one model for each desired emission, which is to be analyzed in terms of operating behavior.
  • Means of comparison means 14 may now the performance of the test confi guration 1 with the performance of the basic configuration 2 are compared 104.
  • values or value profiles of one or more sizes ver equalized which both in the part of the first record X y var var and the second data set y base are included.
  • Fig. 1 this is shown by the two curves of a diagram.
  • One curve shows the course of a variable y base which is assigned to a vehicle of the basic configuration 2 'under changed operating conditions. The vehicle itself is accordingly also provided with a roof.
  • This curve y base is compared with a curve y var of a variable y var , which was measured during the test drive of the configuration 1 to be tested.
  • a characteristic variable can be determined 105 and output 106 via an interface 16, in particular a data interface or user interface.
  • the vehicle type can be optimized 107 on the basis of this parameter. This is shown in FIG. 2 by an arrow which indicates the information that flows from the comparison into a configuration change of a configuration 1 to be tested again.
  • a turbo-charged gasoline vehicle with direct injection according to the Euro 6b emissions standard is to be tested under RDE conditions with various exhaust gas treatment systems including a modern three-way catalytic converter TWC and catalyzed gasoline particle filter cGPF. On the basis of this test, the effects of various configurations on the emission are to be analyzed and evaluated.
  • the data sets X base , X var for the evaluation are collected with the same vehicle with two different exhaust gas aftertreatment systems.
  • the basic configuration is a new three-way catalytic converter Fresh TWC without petrol particle filter cGPF.
  • the configuration to be tested is a so-called end-of-life configuration EoL, i.e. at the end of the service life, with a three-way catalytic converter TWC and a catalyzed gasoline particulate filter cGPF.
  • the test drives are carried out with different drivers and measurement data is generated over a long period of around three months. Four different test tracks are used.
  • Aggregation function S ⁇ maximum, mean, median, minimum, standard deviation, variance, sum, length ⁇
  • Random Forrest regression is used in the present example for each emission based on the measurement data from the test drives. Random Forrest algorithms can handle a large number of features in terms of a small number of samples and are robust.
  • the vehicle model After the vehicle model has been created, it is used for a conclusion-based configuration comparison.
  • the conclusion-based configuration comparison enables the comparison of emissions between a basic configuration and a configuration to be tested, taking into account the ambient conditions, even if only one trip is carried out with the configuration to be tested. This is a great advantage over approaches that attempt to eliminate environmental conditions through a sufficient number of RDE test drives.
  • the diagram in FIG. 6 shows simulated effects of the configuration to be tested EoL cGPF in comparison to the basic configuration Fresh TWC.
  • the nitrogen oxide emissions NO x show a reduction in the configuration to be tested cGPF.
  • the configuration to be tested cGPF shows a significant reduction in the particle number PN due to the presence of the particle filter, which was to be expected.
  • FIG. 6 it is an essential advantage that the test drives of the basic configuration are transformed by the application of the vehicle model 0 base to the same time base or route base as the test drive of the configuration to be tested. This enables a direct value advantage.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Geometry (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Computer Hardware Design (AREA)
  • Mathematical Optimization (AREA)
  • Mathematical Analysis (AREA)
  • Computational Mathematics (AREA)
  • Pure & Applied Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Artificial Intelligence (AREA)
  • Aviation & Aerospace Engineering (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Chemical & Material Sciences (AREA)
  • Combustion & Propulsion (AREA)
  • Mechanical Engineering (AREA)
  • Health & Medical Sciences (AREA)
  • Combined Controls Of Internal Combustion Engines (AREA)

Abstract

L'invention concerne un procédé assisté par ordinateur et un système correspondant d'analyse et/ou d'optimisation d'une configuration d'un type de véhicule sur la base d'un mode de test RDE, le procédé comprenant les étapes de travail suivantes : l'enregistrement d'un premier ensemble de données de variables de fonctionnement mesurées d'une configuration à tester pour un véhicule du type de véhicule et de variables d'environnement en fonction d'une distance parcourue et/ou d'une période dans le passé pendant un mode de test RDE de la configuration à tester, une partie du premier ensemble de données caractérisant un comportement de fonctionnement de la configuration à tester dans des conditions de fonctionnement spécifiques ; la simulation d'un comportement de fonctionnement d'une configuration de base d'un véhicule du type véhicule au moyen d'un modèle de véhicule de la configuration de base, la simulation impliquant une ou plusieurs variables du premier ensemble de données étant des variables d'entrée du modèle de véhicule et un second ensemble de données d'au moins une variable de fonctionnement de la configuration de base étant généré, et le second ensemble de données caractérisant un comportement de fonctionnement de la configuration de base dans les conditions de fonctionnement spécifiques ; et la comparaison du comportement de fonctionnement de la configuration à tester avec le comportement de fonctionnement de la configuration de base sur la base de la partie du premier ensemble de données et du second ensemble de données.
PCT/AT2020/060399 2019-11-12 2020-11-12 Procédé et système d'analyse et/ou d'optimisation d'une configuration d'un type de véhicule WO2021092639A1 (fr)

Priority Applications (1)

Application Number Priority Date Filing Date Title
DE112020004941.8T DE112020004941A5 (de) 2019-11-12 2020-11-12 Verfahren und System zum Analysieren und/oder Optimieren einer Konfiguration einer Fahrzeuggattung

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
ATA50965/2019 2019-11-12
ATA50965/2019A AT523093A1 (de) 2019-11-12 2019-11-12 Verfahren und System zum Analysieren und/oder Optimieren einer Konfiguration einer Fahrzeuggattung

Publications (1)

Publication Number Publication Date
WO2021092639A1 true WO2021092639A1 (fr) 2021-05-20

Family

ID=73835275

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/AT2020/060399 WO2021092639A1 (fr) 2019-11-12 2020-11-12 Procédé et système d'analyse et/ou d'optimisation d'une configuration d'un type de véhicule

Country Status (3)

Country Link
AT (1) AT523093A1 (fr)
DE (1) DE112020004941A5 (fr)
WO (1) WO2021092639A1 (fr)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114897271A (zh) * 2022-06-14 2022-08-12 郑州轻工业大学 数字孪生环境下基于故障传播的中央空调预测性维护方法
CN115263505A (zh) * 2022-08-01 2022-11-01 天津布尔科技有限公司 一种车辆nox排放预测方法、装置及存储介质
CN116424332A (zh) * 2023-04-10 2023-07-14 重庆大学 深度强化学习型混合动力汽车能量管理策略增强更新方法
CN116971881A (zh) * 2023-09-20 2023-10-31 无锡映诺汽车科技有限公司 一种基于数字孪生技术的内燃发动机管理方法及系统

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2016170063A1 (fr) * 2015-04-21 2016-10-27 Avl List Gmbh Procédé et dispositif d'optimisation d'un dispositif technique sur la base d'un modèle
DE102017218476A1 (de) * 2017-10-16 2019-04-18 Robert Bosch Gmbh Verfahren und Vorrichtung zum Ermitteln von Emissionen
DE102018201933A1 (de) * 2018-02-07 2019-08-08 Avl List Gmbh Verfahren und System zur Analyse wenigstens einer Einrichtung einer Einheit, welche eine Mehrzahl an verschiedenen Einrichtungen aufweist

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2016170063A1 (fr) * 2015-04-21 2016-10-27 Avl List Gmbh Procédé et dispositif d'optimisation d'un dispositif technique sur la base d'un modèle
DE102017218476A1 (de) * 2017-10-16 2019-04-18 Robert Bosch Gmbh Verfahren und Vorrichtung zum Ermitteln von Emissionen
DE102018201933A1 (de) * 2018-02-07 2019-08-08 Avl List Gmbh Verfahren und System zur Analyse wenigstens einer Einrichtung einer Einheit, welche eine Mehrzahl an verschiedenen Einrichtungen aufweist
WO2019153026A1 (fr) 2018-02-07 2019-08-15 Avl List Gmbh Procédé et système d'analyse d'au moins un dispositif d'une unité comportant une pluralité de dispositifs différents

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
G. KENDALL: "The treatment of ties in ranking problems", BIOMETRIKA, vol. 33, no. 3, 1945, pages 239 - 251
K. L. LAWRENCE: "A Concordance Correlation Coefficient to Evaluate Reproducibility", BIOMETRICS, vol. 45, no. 1, 1989, pages 255 - 268
RAND R.: "Lehrbuch Wilcox", 2003, ELSEVIER, article "Applying contemporary Statistical techniques"
Y. BENJAMINID. YEKUTIELI: "The Control of the False Discovery Rate in Multiple Testing ander Dependency", THE ANNALS OF STATISTICS, vol. 29, no. 4, 2001, pages 1165 - 1188, XP055720957, DOI: 10.1214/aos/1013699998

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114897271A (zh) * 2022-06-14 2022-08-12 郑州轻工业大学 数字孪生环境下基于故障传播的中央空调预测性维护方法
CN115263505A (zh) * 2022-08-01 2022-11-01 天津布尔科技有限公司 一种车辆nox排放预测方法、装置及存储介质
CN116424332A (zh) * 2023-04-10 2023-07-14 重庆大学 深度强化学习型混合动力汽车能量管理策略增强更新方法
CN116424332B (zh) * 2023-04-10 2023-11-21 重庆大学 深度强化学习型混合动力汽车能量管理策略增强更新方法
CN116971881A (zh) * 2023-09-20 2023-10-31 无锡映诺汽车科技有限公司 一种基于数字孪生技术的内燃发动机管理方法及系统
CN116971881B (zh) * 2023-09-20 2023-12-05 无锡映诺汽车科技有限公司 一种基于数字孪生技术的内燃发动机管理方法及系统

Also Published As

Publication number Publication date
AT523093A1 (de) 2021-05-15
DE112020004941A5 (de) 2022-07-28

Similar Documents

Publication Publication Date Title
WO2021092639A1 (fr) Procédé et système d'analyse et/ou d'optimisation d'une configuration d'un type de véhicule
EP3374748B1 (fr) Méthode de génération d'une séquence de test
EP0846945A2 (fr) Procédé d'analyse du comportement de conduite de véhicules
AT518677B1 (de) Verfahren zum Erstellen eines Analysedatensatzes
EP3540403B1 (fr) Procédé de mise en oeuvre d'un essai d'un spécimen d'essai
DE112008001654T5 (de) System und Verfahren für virtuelle Sensoren
DE102017218476A1 (de) Verfahren und Vorrichtung zum Ermitteln von Emissionen
EP3268713B1 (fr) Procédé de réalisation d'un ensemble de modèles permettant d'étalonner un appareil de commande
AT520827B1 (de) Verfahren zum Bestimmen eines Fahrzeugparameters eines Fahrzeugdatensatzes eines Fahrzeugs und Verwendung des Fahrzeugparameters an einem Prüfstand
DE102019114378A1 (de) Verfahren und system zur vorhersage des systemstatus
DE102008057494A1 (de) Verfahren zum Betrieb eines Verbrennungsmotors
EP2088486A1 (fr) Procédé de mesure d'un système réel dynamique non linéaire selon un plan d'expériences
WO2018177526A1 (fr) Analyse de robustesse sur des véhicules
DE4434197C2 (de) Diagnosetechnik für den Betrieb eines Abgassauerstoffsensors
DE102017106943A1 (de) Verfahren und Anordnung zur Simulation von Fahrversuchen
AT523850A4 (de) Computergestütztes Verfahren und Vorrichtung zur wahrscheinlichkeitsbasierten Geschwindigkeitsprognose für Fahrzeuge
DE102008004218B4 (de) Verfahren zur Bestimmung der dynamischen Rußemission
AT506339B1 (de) Verfahren zur dynamischen totzeitkompensation der abgaskonzentration
EP3942372B1 (fr) Procédé de validation de paramètres de système d'un système énergétique, procédé pour faire fonctionner un système énergétique et système de gestion d'énergie pour un système énergétique
EP1246034B1 (fr) Procédé pour l'analyse et l'estimation de valeurs de mesure d'un système d'essai
EP3851656B1 (fr) Procédé et dispositif d'évaluation des valeurs mesurées déterminées lors de la conduite pratique d'un véhicule
AT525949B1 (de) Verfahren zum Bestimmen einer Konzentration eines Gases in einem Gaspfad einer Brennkraftmaschine
DE102017205437A1 (de) Robustheitsanalyse bei Fahrzeugen
DE102021115103B3 (de) Verfahren, Vorrichtung, Fahrzeug und Computerprogramm zum Modellieren und Überwachen eines Aufheizverhaltens eines Fahrzeugbauteils
DE102023000357B3 (de) Verfahren zum Erzeugen von Testdaten für eine Simulation eines Assistenzsystems eines zumindest teilweise assistiert betriebenen Kraftfahrzeugs, Computerprogrammprodukt, computerlesbares Speichermedium sowie elektronische Recheneinrichtung

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20824425

Country of ref document: EP

Kind code of ref document: A1

REG Reference to national code

Ref country code: DE

Ref legal event code: R225

Ref document number: 112020004941

Country of ref document: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20824425

Country of ref document: EP

Kind code of ref document: A1