Identification of Polytopic Models for a Linear Parameter-varying System Performed on a Vehicle

This paper deals with the parameter identification of continuous time polytopic models for a linear parameter-varying system (LPV). A continuous-time nonlinear identification approach is presented, a mix between a local approach and a global one is introduced in order to identify a LPV model for the lateral comportement of a vehicle. The proposed approach is based on the prediction error method for LTI systems, which is modified to take into account polytopic models and regularization terms. Using experimental data, different parameter-varying structures, explaining the lateral behavior of the vehicle, were identified by the proposed method considering the velocity as the scheduling parameter.