Abstract
We propose a dynamic model validation procedure that provides accurate parameter settings for minimal output errors between simulation models and real experiments. Optimal operations for setting parameters are developed to maximize the effects of specific model parameters while minimizing their interactions.
To manage the high cost of simulating complex systems, the procedure has three main features: optimal excitation based on global sensitivity analysis (GSA) via metamodel techniques, parameter estimation with a polynomial-chaos-based Kalman filter, and model validation through hypothesis testing. An illustrative mathematical model demonstrates the detailed process, and the method is applied to a vehicle dynamics case with a composite maneuver that excites unknown parameters such as inertial properties and tire-model coefficients; the unknown parameters are successfully estimated within a 95% credible interval.
Keywords: Model Validation, Excitation, Parameter Estimation, Maneuver Design, Global Sensitivity Analysis, Kriging, DACE, Kalman Filter