A novel global sensitivity analysis on the observation accuracy of the coupled vehicle model

This paper proposes a novel 2-step global sensitivity analysis algorithm to provide an in-depth sensitivity analysis of the vehicle parameters on the system responses. A 9 degree-of-freedom nonlinear vertical–lateral coupled vehicle model is developed, and 12 parameters along with 7 system responses are selected for the sensitivity analysis. In order to reduce the computational effort, the proposed 2-step algorithm calculates the elementary effects and selects 7 influential parameters from these 12, and then uses the Sobol' method to obtain the global sensitivity indexes of these influential parameters. An unscented Kalman filter is finally designed to show the effect and importance of the selected sensitive parameters on the observation accuracy. Simulations with different vehicle types, velocities and tyre models have verified that the proposed algorithm can accurately describe the relationship between the vehicle parameters and system responses, which is of directive significance to improve vehicle controller and observer performances.