A Composite Linear and Nonlinear Approach to Full-Vehicle Simulator Control 2005-01-0937

This paper presents an approach to full-vehicle simulator control which accounts for nonlinearities in a vehicle/simulator system. The control scheme presented is based on the estimation of the system inverse dynamics. A composite linear/nonlinear approach to inverse system identification (SYS-ID) is presented. The linear portion of the SYS-ID uses time-domain methods to estimate the impulse response of the inverse system in a least squares sense. These results are then extended by using the regularized approach to least squares estimation. The nonlinear part uses the support vector machine to approximate the nonlinear deviations from the linear model. Two approaches to using this composite model are presented. Examples of the linear SYS-ID techniques are shown for a 2×2 system.