Optimal Design of Experiments for Estimating Parameters of a Vehicle Dynamics Simulation Model

Abstract

The calibration of complex simulation models for vehicle component and controller development usually relies on numerical methods. In this contribution, a two-level optimization scheme for estimating unknown model parameters in a commercial real-time capable vehicle dynamics program is proposed. In order to increase the reliability of the model coefficients estimated from reference data, the measuring test is improved by methods for the optimal design of experiments. Specifically, the control variables of the experimental setup are adjusted in such a way as to maximize the sensitivity of the parameters in demand with respect to the objective function. The numerical results show that this two-level optimization scheme is capable of estimating the parameters of a multibody suspension model.