Metric-Based Sampling of Steer-by-Wire Simulation Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods fail to systematically select representative product samples for serial products, particularly steering systems, to characterize model uncertainties due to manufacturing tolerances and aging, which is crucial for steer-by-wire and highly automated driving systems, and lack methods for establishing models with known uncertainties for controller design.
Innovation Solution
A computer-implemented method using dissimilarity metrics (gap, v-gap, and L2 metrics) to determine a predetermined number of parameterized simulation models, ensuring representative samples with quantifiable residual uncertainty, enabling systematic characterization of model uncertainties and controller design.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional expert-based sample selection methods are used, then the selection process is simple and quick, but the selected samples do not adequately represent the entire operational design domain and model uncertainties cannot be systematically characterized
Solution Approach 1:
The method transforms the sample selection problem into a parameter optimization problem by defining representativeness through mathematical metrics (gap metric, v-gap metric, L2 metric) that quantify the distance between sample models and the overall product behavior across the operational design domain. This allows systematic selection based on measurable parameters rather than expert judgment.
Solution Approach 2:
The method replaces the mechanical/expert-based sample selection process with an automated computational approach using dissimilarity metrics and optimization algorithms. The computer system automatically determines representative samples by minimizing mathematical distance measures, eliminating the need for manual expert selection while improving representativeness.
2Measurement precision
If simulation models include all product parameters to accurately represent real behavior, then model accuracy improves, but model complexity and computational requirements increase significantly
Solution Approach 1:
The method extracts only the essential parameters needed for accurate simulation by identifying which parameters contribute most to the gap metric and v-gap metric calculations. This allows the simulation model to focus on critical parameters while omitting less influential ones, maintaining accuracy while reducing complexity.
Solution Approach 2:
The method uses a limited set of representative samples (partial characterization) rather than requiring complete characterization of all possible product variations. By strategically selecting a small number of representative samples that capture the essential uncertainty ranges, the method achieves sufficient model accuracy without the computational burden of exhaustive parameter exploration.
3Reliability
If strict standardized requirements for product approval are implemented for steer-by-wire and highly automated driving systems, then safety and reliability improve, but the number and complexity of real testing and trials increase exceptionally
Solution Approach 1:
The method creates virtual copies of physical product samples through parameterized simulation models. These digital twins replicate the behavior of real steering systems across the operational design domain, allowing approval processes to evaluate multiple virtual variants without requiring proportional increases in physical testing. The simulation models serve as substitutes for extensive real-world trials.
Solution Approach 2:
The method performs preliminary characterization of model uncertainties and validation of simulation models before actual product approval testing. By systematically determining representative samples and characterizing uncertainties in advance, the method prepares comprehensive validation data that can be used to justify approval decisions, reducing the need for extensive post-development testing.
Data Source
AI summary
A computer-implemented method for determining representative parameterized simulation models of a parameterizable analytical simulation model for a product, in particular a steer-by-wire steering system and/or a steering system for highly automated driving, is disclosed. The method includes determining a predetermined number of parameterized simulation models based on a dissimilarity metric, wherein the dissimilarity metric defines a distance between two respective parameterized simulation models, and wherein a plurality of representative parameterized simulation models results.