SYSTEMATIC SELECTION OF REPRESENTATIVE SAMPLES FOR SERIES PRODUCTS

Abstract

A computer-implemented method for determining representative parameterized simulation models of a parameterizable simulation model for a product, in particular a steer-by-wire steering system and / or a steering system for highly automated driving, is disclosed, comprising: calculating a dissimilarity metric on each pair of a plurality of pairs of parameterized simulation models, whereby each pair yields a distance and thereby results in a plurality of distances, optionally wherein the dissimilarity metric is based on a gap metric, a v-gap metric and / or an L2 metric; and selecting a predetermined number of parameterized simulation models based on the plurality of distances, resulting in a plurality of representative parameterized simulation models.

Description

State of the art

[0001] Mass-produced products (especially high-volume production) such as steering systems are subject to variations in product parameters such as friction, elasticity, and / or inertia due to manufacturing tolerances and inaccuracies. Furthermore, mass-produced products are subject to additional parameter variations due to aging, for example, from wear and / or environmental influences. All the ranges and combinations of parameter variations that occur in reality for a mass-produced product constitute its so-called "operational design domain" (ODD).

[0002] Typically, a simulation model (or simply model) of a product only incorporates its essential or relevant properties, which leads to deviations between real-world and modeled behavior due to simplifications. Parameter variation in the ODD (Optical Design Data) results in an additional deviation for each individual production product between its real-world behavior and the corresponding simulation model, which is often focused on representing a product with nominal parameters. This simulation model, including the model deviations and uncertainties across the entire ODD due to simplifications and variations, often forms the basis for developing product control and / or simulation-based product release. Since a complete characterization of the model uncertainties of a production product is currently usually too complex, the simulation models are compared with a few real-world product samples (or simply samples).Typically, product samples are selected as follows: . - Samples with minimum, maximum and / or nominal product parameters that are considered most important by experts. - Samples with product parameters defined by experts.

[0003] Therefore, it is not guaranteed that the selected product samples adequately represent the entire ODD and thus enable a characterization of the model uncertainties in the ODD due to simplification and variation.

[0004] One problem to be solved, underlying the disclosure, could be, for example, providing a method for selecting sufficiently representative product samples for a mass-produced product. Another problem to be solved could be, for example, enabling a characterization of the model uncertainties in the ODD that are due to simplification and variation.

[0005] Compared to traditional steering systems, steer-by-wire (SbW) systems and / or steering systems for highly automated driving (HAD) are subject to stricter regulatory requirements for product release. To prevent the actual testing and validation effort from increasing disproportionately due to these stricter release requirements for mass-produced SbW and HAD steering systems, the industry is focusing on simulation-based release processes. A validated and / or verified simulation model of the steering system with known model uncertainties is essential for such simulation-based release.

[0006] In principle, numerous methods are known for validating and / or verifying various aspects of a simulation model or the entire model. However, there is currently no method that allows for a systematic or complete characterization of the model uncertainties in the ODD (Optimized Design Data) of a (mass-produced) product due to simplification and variation. In particular, no method exists for the systematic or optimal selection of, for example, a few product samples that are representative of the entire ODD with a quantifiable residual uncertainty, thereby enabling a systematic characterization of the model uncertainties in the ODD. Furthermore, steering systems are subject to non-negligible parameter variation and are usually operated in a closed-loop control system.

[0007] Therefore, robust controllers for steering systems are often designed using established methods in control engineering. However, these design methods require a model with known uncertainties for the product. Currently, there is no method to establish one or more models with known uncertainties for controller design that also possess a known representativeness range in the ODD and are collectively representative of the entire ODD.

[0008] Therefore, another problem to be solved, which underlies the disclosure, can also be seen as the need to establish one or more models with known uncertainties for a controller design, which also have a known representativeness range in the ODD and together are representative for the entire ODD.

[0009] In systems theory, various metrics are known that quantify the dissimilarity of two systems—systems and products can subsequently be considered equivalent—and thus compare them. The gap metric, v-gap metric, and L2 metric are explained below.

[0010] The gap metric quantifies the dissimilarity of the open-loop input / output behavior of two systems P1 and P2 with respect to their stability and performance characteristics in closed-loop operation using a scalar in the real interval [0, 1]. A metric result close to 0 indicates that both systems are very similar and any controller stabilizing P1 will also stabilize system P2 with similar controlled performance. A metric result of 0 means that the systems P1 and P2 behave exactly identically. Conversely, a metric result close to or at 1 indicates that systems P1 and P2 are very dissimilar. Furthermore, the gap metric allows for statements regarding the robust stability of closed-loop control systems with model uncertainties. An explicit controller design is necessary for evaluating the gap metric.Details on the definition and properties of the gap metric are described in Chapter 17 of the book “Essentials of Robust Control”, Kemin Zhou and John C. Doyle, 1st edition, Pearson, 1997, ISBN: 9780135258332.

[0011] The system-theoretical statements and implications of the v-gap metric are very similar to those of the gap metric; however, the two metrics are fundamentally defined differently. Evaluating the v-gap metric does not require controller design but does necessitate an analysis of the number of turns in the systems P1 and P2 being compared. Details regarding the definition and properties of the v-gap metric are described in Chapter 17 of the book "Essentials of Robust Control" by Kemin Zhou and John C. Doyle, 1st edition, Pearson, 1997, ISBN: 9780135258332, or in the publication "Frequency domain uncertainty and the graph topology" by Glenn Vinnicombe, IEEE Transactions on Automatic Control, vol. 38, no. 9, pp. 1371-1383, Sept. 1993, DOI: 10.1109 / 9.237648.

[0012] The definition of the L2 metric corresponds to the v-gap metric without the analysis of the number of turns. Therefore, the fundamental statements and implications of both metrics are similar, but the L2 metric has less theoretical power. Details regarding the definition and properties of the L2 metric are described in Chapter 17 of the book "Essentials of Robust Control" by Kemin Zhou and John C. Doyle, 1st edition, Pearson, 1997, ISBN: 9780135258332.

[0013] All three presented dissimilarity metrics are known to possess the following properties: - The metrics quantify the dissimilarity of two systems with respect to stability and performance characteristics in controlled operation based on the uncontrolled input / output behavior. - Using the metric results, it can be decided whether two systems P1 and P2 are sufficiently similar so that P1 can be considered representative of P2. - The metric results can also be interpreted as (system-theoretical) distances between the compared systems. - For all systems to be compared: L2 result ≤ v-gap result ≤ gap result. Disclosure of the invention

[0014] A first general aspect of the present disclosure relates to a computer-implemented method for determining representative parameterized simulation models of a parameterizable simulation model for a product, in particular a steer-by-wire steering system and / or a steering system for highly automated driving.

[0015] The procedure involves calculating a dissimilarity metric for each pair of parameterized simulation models, resulting in a distance for each pair and thus a multitude of distances. The dissimilarity metric can be based, for example, on a gap metric, a v-gap metric, and / or an L2 metric.

[0016] The procedure further includes selecting a predetermined number of parameterized simulation models based on the multitude of distances, resulting in a multitude of representative parameterized simulation models.

[0017] A second general aspect of the present disclosure relates to a computer system designed to execute the computer-implemented method for determining representative parameterized simulation models of a parameterizable simulation model for a product according to the first general aspect (or an embodiment thereof).

[0018] A third general aspect of the present disclosure relates to a computer program designed to execute the computer-implemented method for determining representative parameterized simulation models of a parameterizable simulation model for a product according to the first general aspect (or an embodiment thereof).

[0019] A fourth general aspect of the present disclosure relates to a computer-readable medium or signal that stores and / or contains the computer program according to the third general aspect (or an embodiment thereof).

[0020] The method proposed here according to the first general aspect (or an embodiment thereof) allows a few product samples to be systematically selected based on a model-based criterion, such that they represent the entire ODD of a serial product with a quantifiable residual uncertainty.

[0021] In particular, the following advantages compared to the prior art can be achieved by the method proposed here according to the first general aspect (or an embodiment thereof): - Systematic approach to selecting product samples that are representative of the entire ODD; - Quantifiable residual uncertainty of the product samples with respect to the total ODD; - Systematic characterizability of the model uncertainties caused by simplification and dispersion across the entire ODD; - Calculable representativeness ranges of the respective product samples in the entire ODD; - Systematically derivable requirement for the manufacturing accuracy of the product samples; - Systematically derivable requirements for the design of a robust product regulation; and / or - Optional search for product parameters or combinations that significantly alter product behavior.

[0022] The method proposed here according to the first general aspect (or an embodiment thereof) can be used in the development of real products - e.g., in steer-by-wire (SbW) steering systems - in the design phase and / or in system development (i.e., after the design phase).

[0023] For example, in the design phase, the method can be used to systematically select (e.g., a few) product samples that are representative of all parameter ranges and combinations that occur in reality for a mass-produced product (i.e., for the entire ODD). Furthermore, intermediate results of the method can be used to identify individual product parameters or combinations that have a particularly strong impact on product behavior.

[0024] Based on the representative product samples, the uncertainties arising from parameter variation and modeling simplifications between the actual product behavior and its modeled behavior across the entire ODD can be systematically characterized during further system development. The simulation model, including the characterized model uncertainties, can then be used for developing a product control system and / or for simulation-based product release. Additionally, the method proposed here can be used, for example, to systematically establish any number of controller design models (i.e., one or more representative models for a given controller design), including their associated model uncertainties. Each model has a known representativeness range within the ODD and together they are representative of the entire ODD.These controller design models can be used, for example, for the systematic development of a gain scheduling control for the product, taking into account the known model uncertainties and representativeness ranges.

[0025] The method proposed here, according to the first general aspect (or an embodiment thereof), can be applied, in particular, within the framework of a simulation-based release process for SbW and HAD steering systems. This requires the selection of product samples representative of the entire ODD for the validation and / or verification of simulation models for, e.g., steering systems. Furthermore, the method proposed here can, in principle, also be used for model validation and / or verification and / or within the context of simulation-based release for other (large-scale) production products.

[0026] The method proposed here according to the first general aspect (or an embodiment thereof) can be carried out wholly or partially numerically. This is advantageous because it does not depend on the parameterizable simulation model being available in analytical form. Brief description of the characters Fig. Figure 1 schematically illustrates exemplary embodiments of a computer-implemented method for determining representative parameterized simulation models of a parameterizable simulation model for a product, in particular a steer-by-wire steering system and / or a steering system for highly automated driving. Fig. Figure 2 illustrates an exemplary embodiment of the computer-implemented method based on an adjacency matrix. Fig. Figure 3a illustrates an exemplary ODD with a variety of parameterized simulation models, of which three are selected as representative parameterized simulation models. Fig. 3b illustrates the exemplary ODD from Fig. 3a with the three selected representative parameterized simulation models. Detailed description

[0027] The method 100 proposed in this disclosure can be aimed at selecting sufficiently representative product samples for a series product. Alternatively or additionally, the method 100 can also be aimed at enabling a characterization of the model uncertainties in the ODD caused by simplification and variation. Alternatively or additionally, the method 100 can also be aimed at establishing one or more models with known uncertainties for a controller design, which also have a known representativeness range in the ODD and together are representative for the entire ODD.

[0028] The following describes the systematic and model-based selection of representative product samples.

[0029] Method 100 can initially be aimed at determining representative parameterized simulation models of a parameterizable simulation model. Alternatively or additionally, method 100 can be aimed at determining representative parameter samples in the ODD. Alternatively or additionally, method 100 can be aimed at selecting the representative product samples.

[0030] The selection of representative product samples can be based on the specific representative parameterized simulation models of the parameterizable simulation model and / or on the representative parameter samples in the ODD.

[0031] Method 100 is wholly or partially numerical and can therefore be used even if the parameterizable simulation model is not available in analytical form.

[0032] First, a computer-implemented method 100 for determining representative parameterized simulation models of a parameterizable simulation model for a product is disclosed. The product can, in particular, be a mass-produced product, i.e., manufactured in series. Method 100—although it can also be applied to a non-mass-produced product or a small series—is especially useful when a large number of similar products are to be manufactured, which may nevertheless differ (e.g., due to production and / or material differences). The number of similar products can, for example, comprise > 100 products per year, > 500 products per year, or > 106 products per year.

[0033] The product could, for example, be a steer-by-wire steering system. Alternatively or additionally, the product could be a steering system for automated, especially highly automated, driving.

[0034] The parameterizable simulation model can be analytical, but it doesn't have to be. For example, in Fig. As schematically illustrated, the procedure comprises calculating a dissimilarity metric on each pair of a multitude of pairs of (generally different) parameterized simulation models, resulting in a distance for each pair and thus a multitude of distances.

[0035] The dissimilarity metric can be based on a gap metric, a v-gap metric, and / or an L2 metric. Specifically, the dissimilarity metric can be the gap metric, the v-gap metric, or the L2 metric. Alternatively, the dissimilarity metric can be based on a combination of the gap metric, the v-gap metric, and / or the L2 metric. The dissimilarity metric can output a quantitative measure of the dissimilarity of a pair of parameterized simulation models; that is, a quantitative measure of how similar or dissimilar the two parameterized simulation models of the pair are. In this respect, the dissimilarity metric could also be called a similarity or comparison metric. The quantitative measure output by the dissimilarity metric can be referred to as distance. If the two parameterized simulation models of a pair are similar, the distance can be small. In particular, if the two parameterized simulation models of a pair are identical (i.e.,If the two parameterized simulation models of a pair are maximally similar, the distance can be zero. Conversely, if the two parameterized simulation models of a pair are not similar, the distance can be high.

[0036] For example, in Fig. As schematically illustrated in Figure 1, the procedure further comprises selecting a predetermined number of parameterized simulation models based on the multitude of distances, resulting in a multitude of representative parameterized simulation models. An exemplary result of such a selection process is shown in Figure 1. Fig. 3a-b shown.

[0037] The multitude of representative parameterized simulation models can be defined such that the distances of each parameterized simulation model (in the ODD) to the nearest representative parameterized simulation model (also in the ODD) with respect to the dissimilarity metric are minimal. For example, these distances can be minimal in a predetermined finite-dimensional norm. It is also conceivable, for instance, to modify the distances before applying the finite-dimensional norm, in particular to weight them, e.g., based on knowledge about the product and / or its ODD that is already available at the time of execution of procedure 100.

[0038] Here, the minimum distances, in particular the minimum distance in the finite-dimensional norm, can be determined. The minimization can be performed in a mathematically rigorous or an approximate sense. For example, it may suffice to find a local minimum instead of a global minimum. Thus, the selection of the predetermined number of representative parameterized simulation models based on the multitude of distances can be exact or approximate.

[0039] The predetermined finite-dimensional norm can, for example, be a p-norm for any p in the interval [1, +Inf], where +Inf denotes positive infinity. In particular, the p-norm can be a sum norm (i.e., p=1). In this case, the sum of all distances can be minimal. In another example, the p-norm can be a Euclidean norm (i.e., p=2). In this case, the Euclidean distance can be minimal. In yet another example, the p-norm can be a maximum norm (p=Inf). In the case of the maximum norm, the multitude of representative parameterized simulation models can be defined such that the maximum of all distances of each parameterized simulation model to its respective nearest representative parameterized simulation model is minimal.

[0040] For example, in Fig. As schematically illustrated as an option, the procedure can further include determining a multitude of parameter samples in an Operational Design Domain (ODD) for the product. Each parameter sample in the ODD can comprise one or more parameters of the parameterizable simulation model. The ODD can be defined by a numerical and / or analytical description. Determining the multitude of parameter samples in the ODD can, for example, be done by ensuring that the ODD is covered sufficiently uniformly. Such sufficiently uniform coverage can be achieved, for example, based on pseudorandom numbers. Pseudorandom numbers (e.g., via the Mersenne Twister) are typically uniformly distributed. Alternatively or additionally, such sufficiently uniform coverage can be based on Latin Hypercube Sampling. Alternatively or additionally, such sufficiently uniform coverage can be based on a Sobol sequence.In particular, such sufficiently uniform coverage can be based on a combination of pseudorandom numbers, Latin hypercube sampling, and / or a Sobol sequence. By already assuming sufficiently uniform coverage, more representative parameterized simulation models of the parameterizable simulation model can be selected more efficiently.140

[0041] Step 110 is described in the exemplary embodiment in Fig. 2 referred to as “sampling of the ODD”.

[0042] Alternatively, determining 110 of the multitude of parameter samples can only refer to a part of the ODD (e.g. in the case of certain questions or for further investigation of the ODD).

[0043] For example, in Fig. As schematically illustrated in Figure 1, the procedure can further include the creation of parameterized simulation models based on the parameterizable simulation model and the multitude of parameter samples in the ODD. For example, the parameterizable simulation model can be evaluated on one of the parameter samples at a time. Alternatively or additionally, a surrogate model for the parameterizable simulation model can be created for each of the parameter samples.

[0044] The parameterizable simulation model can be analytical, but it doesn't have to be. If the parameterizable simulation model is not analytical (e.g., in the case of a black-box simulation model), it may not be possible to simply evaluate it on each of the parameter samples. In such a case, a surrogate model for the parameterizable simulation model can be created for each of the parameter samples using numerical simulation. Thus, such a surrogate model can also be seen as a parameterized simulation model of the parameterizable simulation model. Therefore, creating the surrogate model for each parameter sample can constitute a parameterization of the parameterizable simulation model. In this respect, the (parameterizable) simulation model itself can be parameterizable, even if it is not in analytical form.

[0045] Step 120 is described in the exemplary embodiment in Fig. 2 is referred to as “determining the PUM for all ODD samples based on PAM”, where PAM is the parameterizable simulation model and PUM is the multitude of parameterized simulation models.

[0046] For example, in Fig. Illustrated schematically and as an option, the procedure 100 can further include determining 141 a measure for the representativeness of the representative parameterized simulation models based on the predetermined number and / or the minimum distances, in particular the minimum distance in the finite-dimensional norm. The measure for representativeness can, for example, include a quantifiable residual uncertainty of the product samples with respect to the ODD.

[0047] As also, for example, in Fig. As illustrated schematically and as an option, procedure 100 can further include adjusting 150, in particular increasing, the predetermined number if the measure of representativeness does not meet a predetermined criterion. The predetermined criterion may be met, for example, if the measure of representativeness is sufficiently large. After adjusting 150, in particular increasing, the predetermined number, procedure 100 can be repeated, for example, until the measure of representativeness meets the predetermined criterion (i.e., until the measure of representativeness is sufficiently large).

[0048] For example, in Fig. As schematically illustrated as an option, the procedure 100 can further include the creation of an (e.g., symmetric) adjacency matrix based on the multitude of distances. In particular, the components of the adjacency matrix can include the distances. The adjacency matrix can make the implementation of the procedure 100 more efficient.

[0049] The formation of the adjacency matrix (131) can still be based on a sequence of parameterized simulation models. This sequence may be irrelevant. For example, this sequence could be the sequence in which the multitude of parameter samples in the ODD are determined.

[0050] Step 131 is described in the exemplary embodiment in Fig. Section 2 is referred to as "calculation of the metric-based adjacency matrix for all PUM pairs," where PUM again denotes the multitude of parameterized simulation models; thus, each PUM pair is a pair of two PUMs. The metric-based adjacency matrix here refers to the adjacency matrix.

[0051] Selecting 140 of the predetermined number of parameterized simulation models based on the multitude of distances (leading to the multitude of representative parameterized simulation models) can involve a full factorial search in the adjacency matrix. Alternatively or additionally, selecting 140 of the predetermined number of parameterized simulation models based on the multitude of distances can involve (e.g., direct or successive) clustering of the adjacency matrix. Alternatively or additionally, selecting 140 of the predetermined number of parameterized simulation models based on the multitude of distances can involve converting the adjacency matrix into equidistant (with respect to the dissimilarity metric) auxiliary points and clustering these auxiliary points.Selecting 140 of the predetermined number of parameterized simulation models based on the multitude of distances can involve a combination of a full factorial search in the adjacency matrix, clustering of the adjacency matrix, and / or conversion of the adjacency matrix into equidistant (with respect to the dissimilarity metric) auxiliary points and clustering of the auxiliary points. The adjacency matrix can be clustered by providing the entire adjacency matrix for clustering (here also referred to as direct clustering). Alternatively or additionally, the adjacency matrix can be clustered successively and partially (i.e., in bite-sized chunks). The latter can be more computationally and / or memory-efficient. The auxiliary points and / or the clustering of the auxiliary points can (individually or in combination) offer numerical advantages.

[0052] Step 140, based on the adjacency matrix, is implemented in the exemplary embodiment in Fig. 2 is referred to as “search for representative PUM and ODD samples based on the adjacency matrix”, where PUM refers to the multitude of parameterized simulation models and the adjacency matrix refers to the adjacency matrix.

[0053] For example, in Fig. 1 schematically and as an option illustrated, the procedure 100 can further determine 160 the respective parameter samples assigned to the representative parameterized simulation models as the representative parameter samples of the parameterizable simulation model.

[0054] As also, for example, in Fig. As schematically illustrated as an option, the procedure can further include determining the product patterns assigned to the representative parameterized simulation models, resulting in a multitude of representative product patterns. Determining the product patterns assigned to the representative parameterized simulation models can, in particular, be based on the representative parameter samples of the parameterizable simulation model. Determining the multitude of representative product patterns can, for example, be done by comparing them with a database of stored parameter samples of real product patterns, especially based on a similarity relation. In the case of unsatisfactory similarity, one or more instructions can be issued on how to produce a sufficiently satisfactory representative product pattern from a real product pattern (e.g., by setup, shimming, etc.).

[0055] Steps 160 and 161 are described in the exemplary embodiment in Fig. Section 2 is referred to as the “definition of the representative patterns based on the representative PUM and ODD points”. Here, the representative PUM refer to the multitude of representative parameterized simulation models, and the ODD points refer to the ODD samples.

[0056] For example, in Fig. As schematically illustrated in Figure 1, the procedure 100 can further include the design 170 of a product control system based on at least one representative parameterized simulation model. The design 170 of the product control system can, in particular, be based on the multitude of representative parameterized simulation models. This enables the design and / or implementation of a robust product control system. In particular, this allows the product control system to be designed to be especially robust and reliable. This, in particular, can increase the product's safety.

[0057] As also, for example, in Fig. As illustrated schematically and as an option, procedure 100 can further include checking 171 one or more product requirements based on at least one representative parameterized simulation model. Checking 171 the one or more product requirements can, in particular, be based on the multitude of representative parameterized simulation models. This allows the product to be designed to be robust and reliable. This also increases the product's safety.

[0058] For example, in Fig. As schematically illustrated as an option, the procedure can further include identifying one or more of the representative parameterized simulation models and / or the representative parameter samples that have a greater influence on the product and / or its behavior (e.g., steering behavior). This can be done, for example, by means of a sensitivity analysis. This can improve the understanding of the product and its behavior. This can be particularly useful in the case of higher-dimensional ODDs, since the ODD can then no longer be directly observed.

[0059] Fig. Figure 2 shows an exemplary embodiment of method 100. The following steps, for example, can be carried out sequentially: - “Sampling of the ODD” - "Determination of PUM for all ODD samples based on PAM" - "Calculation of the metric-based adjacency matrix for all PUM pairs" - "Search for representative PUM and ODD samples using the adjacency matrix" - “Definition of representative patterns based on representative PUM and ODD points”

[0060] Fig. Figure 3a illustrates an exemplary ODD with a multitude of parameterized simulation models. For each of these parameterized simulation models, the corresponding parameter sample is represented as a filled point in the ODD. In this example, the ODD is a two-dimensional product space, meaning it is spanned by two parameters (one parameter in the x-direction, the other in the y-direction), each of which can take values ​​in an interval. In general, the ODD need not be a product space but can be any manifold. The dimension of the ODD can be arbitrary.

[0061] From the multitude of parameterized simulation models or their parameter samples, three were selected in this example as representative parameterized simulation models or their parameter samples in the ODD 140, 160. The latter are each represented by a cross in the ODD. The three representative parameterized simulation models in this example 140, 160 were selected such that the distances (according to the dissimilarity metric) of each parameterized simulation model to its nearest representative parameterized simulation model are minimal.

[0062] For this example, in 3D, i.e., in the z-direction orthogonal to the plane of the sheet, the distance (according to the dissimilarity metric) of each parameter sample in the ODD (i.e., more precisely, its parameterized simulation model) to the nearest representative parameterized simulation model could be shown. In 2D, instead, several isolines are shown with respect to this quantity in the z-direction. For example, the three solid lines form an isoline, i.e., they have the same z-value. In addition, three more isolines—roughly centered on the crosses—are shown. Alongside the isolines, a boundary of influence areas (dashed lines intersecting at a point) of the respective representative parameterized simulation models is drawn.

[0063] Fig. 3b illustrates the exemplary ODD from Fig.3a with the three selected representative parameterized simulation models (more precisely, their parameter samples).

[0064] Furthermore, a computer system is disclosed that is designed to execute the computer-implemented method 100 for determining representative parameterized simulation models of a parameterizable simulation model for a product. The computer system may include a processor and / or main memory.

[0065] Furthermore, a computer program is disclosed that is designed to execute the computer-implemented method 100 for determining representative parameterized simulation models of a parameterizable simulation model into a product. The computer program can be in interpretable or compiled form, for example. It can be loaded (even partially) into a computer's RAM for execution, for example as a bit or byte sequence.

[0066] Furthermore, a computer-readable medium or signal that stores and / or contains the computer program is disclosed. The medium can include, for example, RAM, ROM, EPROM, HDD, SSD, etc., on / in which the signal is stored. QUOTES INCLUDED IN THE DESCRIPTION

[0000] This list of documents cited by the applicant was automatically generated and is included solely for the reader's convenience. The list is not part of the German patent or utility model application. The DPMA accepts no liability for any errors or omissions. Cited non-patent literature

[0000] Chapter 17 of the book “Essentials of Robust Control”, Kemin Zhou and John C. Doyle, 1st edition, Pearson, 1997, ISBN: 9780135258332 [0010, 0012] Essentials of Robust Control", Kemin Zhou and John C. Doyle, 1st edition, Pearson, 1997, ISBN: 9780135258332

[0011] Frequency domain uncertainty and the graph topology“, Glenn Vinnicombe, IEEE Transactions on Automatic Control, vol. 38, no. 9, pp. 1371-1383, Sept. 1993, DOI: 10.1109 / 9.237648

[0011]

Claims

[1] Computer-implemented method (100) for determining representative parameterized simulation models of a parameterizable simulation model of a product, in particular a steer-by-wire steering system and / or a steering system for highly automated driving, comprising: - Calculating (130) a dissimilarity metric on each pair of a plurality of pairs of parameterized simulation models, where each pair results in a distance and thus results in a plurality of distances; optionally where the dissimilarity metric is based on a gap metric, a v-gap metric and / or an L2 metric; - Selecting (140) a predetermined number of parameterized simulation models based on the multitude of distances, resulting in a multitude of representative parameterized simulation models. [2] Method (100) according to claim 1, wherein the plurality of representative parameterized simulation models is defined such that the distances of each parameterized simulation model to the respective nearest representative parameterized simulation model are minimal, in particular in a predetermined finite-dimensional norm. [3] Method (100) according to claim 1 or 2, wherein the predetermined finite-dimensional norm is a p-norm for p in [1, +Inf], in particular a sum norm, a Euclidean norm or a maximum norm. [4] Method (100) according to one of claims 2 or 3, comprising: - Determining (141) a measure for the representativeness of the representative parameterized simulation models based on the predetermined number and / or minimum distances, in particular the minimum distance in the finite-dimensional norm; - optional, Adjust (150), in particular increase, the predetermined number if the measure of representativeness does not meet a predetermined criterion. [5] Method (100) according to any one of the preceding claims, comprising: - Determine (110) a plurality of parameter samples in an Operational Design Domain (ODD) for the product, optionally wherein each parameter sample in the ODD comprises one or more parameters of the parameterizable simulation model; - Forming (120) the parameterized simulation models based on the parameterizable simulation model and the multitude of parameter samples in the ODD, in particular wherein the parameterizable simulation model is evaluated on each of the parameter samples and / or wherein a surrogate model for the parameterizable simulation model is created for each of the parameter samples. [6] Method (100) according to claim 5, wherein the determination (110) of the plurality of parameter samples in the ODD is carried out in such a way that the ODD is covered sufficiently uniformly, in particular based on pseudo-random numbers, on Latin Hypercube Sampling and / or on a Sobol sequence. [7] Method (100) according to claim 5 or 6, comprising: - Determine (160) the respective parameter samples assigned to the representative parameterized simulation models as the representative parameter samples of the parameterizable simulation model. [8] Method (100) according to any one of the preceding claims, comprising: - Forming (131) an adjacency matrix based on the plurality of distances. [9] Method (100) according to claim 8, wherein the selection (140) of the predetermined number of parameterized simulation models based on the plurality of distances: - a full factorial search in the adjacency matrix; - a clustering of the adjacency matrix; and / or - a conversion of the adjacency matrix into equally spaced auxiliary points and clustering of the auxiliary points; includes. [10] Method (100) according to any one of the preceding claims, comprising: - Determining (161) product patterns assigned to the representative parameterized simulation models, resulting in a variety of representative product patterns, in particular based on the representative parameter samples of the parameterizable simulation model. [11] Method (100) according to any one of the preceding claims, comprising: - Design (170) of a control system for the product based on at least one representative parameterized simulation model, optionally based on the multitude of representative parameterized simulation models; and / or - Checking (171) one or more requirements for the product based on at least one representative parameterized simulation model, optionally based on the multitude of representative parameterized simulation models. [12] Method (100) according to any one of the preceding claims, comprising: - Identify (172) one or more of the representative parameterized simulation models and / or the representative parameter samples that have a greater influence on the product. [13] Computer system designed to perform the computer-implemented method (100) for determining representative parameterized simulation models of a parameterizable simulation model for a product according to any of the preceding claims. [14] Computer program designed to perform the computer-implemented method (100) for determining representative parameterized simulation models of a parameterizable simulation model for a product according to any one of claims 1 to 12. [15] Computer-readable medium or signal that stores and / or contains the computer program according to claim 14.

Images