机械臂神经网络变结构控制器

A Neural Network Variable Structure Robot Controller

本文将变结构原理引入网络控制器的鲁棒性设计．结果表明，此类控制器可自动补偿由于不精确学习或系统参数变化而引起的动态特性上的不足．

In this paper,variable structure control methods are introduced to robustify the neural net work controller. Simulation results demonstrate that this scheme can achieve fast and precise robot motion control under the circumstances of load changing and inaccuracy of inverse-dynamics learning.