基于自定义神经网络的EPS助力特性研究

Study of Assist Characteristic of EPS Based on Self-defined Neural Network

对电动助力转向(EPS)系统的助力特性进行了分析,并利用搭建的自定义神经网络模型对助力特性曲线进行预描,在MATLAB中得到的仿真结果是一组非常理想的助力曲线。这表明该自定义网络能有效预测EPS的助力特性,同时也为后期的回正控制策略研究提供了思路。

This paper analyses the assist control strategy of electric power steering system. On the basis of the study, a three-layer self-defined neural network is constructed and trained to forecast the curve of assist characteristics. Through the simulation in MATLAB, the paper gets a set of smooth assist characteristic curves which we are trying to get. The ideal result shows that this self-defined neural network can effectively predict the assist characteristics of electric power-assisted steering system. At the same time, this new neural network model provides a good approach for the further study of return control strategy.