Abstract
Validation in vehicle engineering identifies and quantifies the differences between simulation models and experiment data. In this work we consider these differences — the lack of ability to model uncertainties and to identify unknown parameter values, especially for coupled complex systems such as vehicles.
Effects of unknown model parameters vary under different maneuvers, and the ability to excite a source of uncertainty is the focus of this study. We propose an optimization method that generates a proper maneuver to maximize the sensitivity of uncertain parameters based on global sensitivity analysis (GSA), including sensitivities with respect to individual parameters and coupled effects. Kriging-based metamodels improve the efficiency of the GSA problems with computationally expensive simulations, yielding an optimal excitation maneuver. The accuracy and applicability are assessed via a math model and a practical application on an x-by-wire autonomous tricycle.
Keywords: Excitation, Metamodel, Global Sensitivity Analysis, Uncertainty, Optimization, Vehicle System