A feedback linearization controller combined with a data-driven subspace-based prediction method for vehicle handling stabilization

Abstract

Obtaining precise yaw rate and lateral velocity as well as developing a nonlinear controller becomes more and more essential for improving the vehicle handling performance. Different from traditional methods, a data-driven subspace-based prediction approach is introduced by integrating propagator with predictor-based subspace identification method in this paper. Based on an identifiable vehicle model, the prediction process is validated by standard road tests data. To employ this data-driven prediction method in the vehicle handling stabilization and solve the controlling problem of nonlinear lateral dynamic system, a feedback linearization controller based on the new piecewise tire model is elaborately developed. On account of that the one-step prediction output reduces the time delay between actuator and lateral dynamic response, the subspace-based controller can theoretically improve the vehicle handling performance. By road simulation results, the proposed feedback linearization controller combined with a data-driven subspace-based prediction method greatly enhances the handling performance and provides a more effective technique for both vehicle parameter estimation and handling stabilization.