A Particle Filter Approach for Identifying Tyre Model Parameters From Full-Scale Experimental Tests

Abstract

Mathematical models simulating the handling behavior of passenger cars are extensively used at a design stage for evaluating the effects of new structural solutions or control systems. The main source of uncertainty in this type of models lies in the tyre-road interaction, due high nonlinearity. Proper estimation of tyre model parameters is thus of utter importance to obtain reliable results. A methodology aimed at identifying the Magic Formula-Tyre (MF-Tyre) model coefficients of the tyres of an axle based only on the measurements carried out on board vehicle (vehicle sideslip angle, yaw rate, lateral acceleration, speed and steer angle) during standard handling maneuvers (step-steers, double lane changes, etc.) is presented in this paper. The proposed methodology is based on Particle Filtering (PF) technique. PF may become a serious alternative to classic model-based techniques, such as Kalman filters. Results of the identification procedure were first checked through simulations. Then PF was applied to experimental data collected on an real instrumented passenger-car vehicles.