航空学报 > 2002, Vol. 23 Issue (6): 525-529

基于神经网络的电动加载系统

沈东凯, 华清, 王占林   

  1. 北京航空航天大学自动化科学与电气工程学院 北京 100083
  • 收稿日期:2001-09-10 修回日期:2002-03-02 出版日期:2002-12-25 发布日期:2002-12-25

MOTOR-DRIVEN LOAD SYSTEM BASED ON NEURAL NETWORKS

SHEN Dong-kai,HUA Qing,WANG Zhan-lin   

  1. College of Automation Science and Electrical Engineering, Beijing University of Aeronautics and Astronautics, Beijing 100083, China
  • Received:2001-09-10 Revised:2002-03-02 Online:2002-12-25 Published:2002-12-25

摘要:

针对电动加载系统中多余力矩的干扰 ,提出了基于RBF(径向基函数 )神经网络的新型复合控制策略 ,与传统的BP神经网络相比 ,没有局部最小问题。由于系统非线性和时变性 ,特别在多余力干扰下传统控制方法如PID很难得到满意的控制效果。提出的复合控制方法主要由神经网络PID和前馈补偿器组成 ,通过仿真与试验 ,控制器有效的减少了多余力矩对系统的影响 ,改善了加载系统的动态性能。

关键词: 电动加载, 前馈补偿, RBF神经网络

Abstract:

Aiming at the disturbance of the extraneous force in the motor-driven load system, a new composite control strategy based on Radial Basis Function (RBF) networks is proposed. Compared with the controllers based on conventional BP networks, the presented algorithm is much more efficient without the problem of local minima. The motor driven load system is highly nonlinear and includes delays in the control loop. It is difficult for the traditional control method such as PID to improve the performance, especially under the disturbance of movement,the so called extraneous force problem. The proposed composite control scheme consists of NN PID and feedforward compensator. The experimental result shows that the scheme compensates the extraneous force effectively, and improves the dynamic performance of the load system.

Key words: motor-drive, feedforward compensation, RBF neural network