电液伺服系统的模糊神经网络并行自学习鲁棒控制

Robust Control Based on Fuzzy Neural Network with Collateral Self-learning in Electro-hydraulic Servo System

基于模糊神经网络结构,提出了一种复合式控制方案,解决了传统自适应控制中模型的在线辨识和控制器的在线设计问题。达到了对不确定非线性系统的高精度榆出跟踪;通过引入运行监控器,解决了模糊神经网络实时性差的问题;同时,利用一个鲁棒反馈控制器,来保证模糊神经网络学习初期闭环系统的稳定性。应用到电流力伺服加载系统中,获得满意控制效果。

In this paper, to solve model identificationproblem and on line design of the controller in traditionalcontrol theory, a kind of hybrid control scheme is pre-sented with the T--S fuzzy neural network structure.This control scheme can track the outputs of the uncer-tain and nonlinear systems with high accuracy; using thesupervisor, the poor real--time contorl problem in fuzzyneural network can be over come. Meanwhile, with a ro-bust feedback controller, the stability of closed--loopsystem in initial stages of fuzzy neural nerwork modellearning can also be guaranteed. And the satisfied resultshave been achieved in applications of a electro--hydraulicservo system.