基于滑模的神经元网络自适应控制及其应用(英文)

Neuron Network Adaptive Control Based on Sliding Mode and Its Application

提出一种基于滑模的神经元网络自适应控制方法,并把它应用于液压伺服系统的位置控制.基于滑模学习策略,根据从一优化了的滑模控制所得到的系统输入/输出信号,设计一神经元网络,离线训练该神经元网络的权值,然后综合一简单的自适应环节,得到完整的基于滑模的神经元网络自适应控制.仿真实验结果表明,相对于纯优化的滑模控制而言,所提出的控制方法能使系统具有响应速度快,控制精度高的特点,综合控制效果明显.

A neuron network adaptive control based on sliding mode is developed and applied to the positioning control of a hydraulic servo system. According to the system's input/output data obtained from an optimized slidingmode control, a neuron network control together with an adaptation mechanism is devised and trained offline based on a sliding mode learning strategy, which is used as an adaptive learning algorithm to train the adjustment weights of neuron network. The simulation experiment results applied to a hydraulic servo system show that the learning approach of neuron network exhibits fast convergence property and can be effectively used for online control. The results also show that the system by using the proposed control has better comprehensive properties than that of by using pure optimized sliding mode control.