Torque Vectoring Predictive Control of a Four In-Wheel Motor Drive Electric Vehicle

Abstract

The recent integration of vehicles with electrified powertrains in the automotive sector provides higher energy efficiency, lower pollution levels and increased controllability. These features have led to an increasing interest in the development of Advanced Driver- Assistance Systems (ADAS) that enhance not only the vehicle dynamic behaviour, but also its efficiency and energy consumption. This master’s thesis presents some contributions to the vehicle modeling, parameter estimation, model predictive control and reference generation applied to electric vehicles, paying particular attention to both model and controller validation, leveraging offline simulations and a real-time driving simulator. The objective of this project is focused on the Nonlinear Model Predictive Controller (NMPC) technique developing torque distribution strategies, specifically Torque Vectoring (TV) for a four-in wheel motor drive electric vehicle. A real-time TV-NMPC algorithm will be implemented, which maximizes the wheels torque usage and distribution to enhance vehicle stability and improve handling capabilities. In order to develop this control system, throughout this thesis the whole process carried out including the implementation requirements and considerations are described in detail. As the NMPC is a model-based approach, a nonlinear vehicle model is proposed. The vehicle model, the estimated parameters and the controller will be validated through the design of open and closed loop driving maneuvers for offline simulations performed in a simulation plant (VI-CarRealTime) and by means of a real-time driving simulator (VI-Grade Compact Simulator) to test the vehicle performance through various dynamic driving conditions.