Controller design for an electric power steering system based on LQR techniques

Source abstract

Purpose This paper aims to present a linear quadratic regulator (LQR) employed to improve performance of an electrical power steering (EPS) system. Design/methodology/approach Generally, EPS is a full electric system having an electrical motor which provides the assist torque on the steering mechanism in order to reduce the workload and to enhance the steering feel of the driver during the steering process. Since the torque sensors are considerably expensive, the authors present a control strategy that eliminates the driver torque sensor by introducing a torque estimator. Three main technical areas are described in this paper. First, the principle and structure of EPS are presented including the dynamic model. Second, LQR and Kalman filter techniques are employed to derive an optimal controller for the EPS system. Finally, the simulations and hardware results are depicted. Findings The combined tools of Matlab/Simulink and dSPACE provide the environment for modelling the controller in software and applying it to the actual hardware via a digital signal processing board based on the DS1401 MicroAutoBox. The controller is evaluated via simulation results, dSPACE hardware results, and verified on vehicle testing data. Originality/value This paper presents a controller design for an EPS system based on the LQR techniques. Within the controller concept shown, elimination of the driver torque sensor offers advantages in terms of both cost and mechanical performance. Simulations and measured data prove the good functionality of the controller proposed.