Electric power steering nonlinear problem based on proportional–integral–derivative parameter self-tuning of back propagation neural network
Yuanyuan Li, Guofei Wu, Liqun Wu, Shaotang Chen
Abstract
Aiming at the problem of nonlinear power steering in the automobile electric power steering system, an advanced control algorithm is required for the practical system. This paper introduces back propagation neural network arbitrary nonlinear approximations to discretize the vehicle’s power assistance characteristic. Steering power is also realized in the whole range of speed, which overcomes the steering blind zone and lays a foundation for the design of subsequent controllers. In addition, considering the nonlinear frictional resistance problem of the electric power steering system, the traditional proportional–integral–derivative remote control algorithm will result in poor dynamic performance or system instability. Therefore, this paper proposes a control algorithm based on back propagation neural network proportional–integral–derivative parameter self-tuning. Using the error between the expected current and the actual motor current, the back propagation neural network algorithm is used to learn and realize the adaptive adjustment of proportional–integral–derivative parameters. Simulation results show that the proposed control system effectively realizes the nonlinear steering assistance in the whole vehicle range speed, eliminates the influence of nonlinear friction in the electric power steering system, and improves the robustness of the control system.