基于RBF神经网络逆系统的机械手解耦控制策略  

A decoupling control strategy of manipulator based on RBF neural network inverse control system

作　　者：刘根水[1] 张绍德[1] 李娟[1] 

机构地区：[1]安徽工业大学电气信息学院,马鞍山243002

出　　处：《自动化与仪器仪表》2008年第6期12-14,共3页Automation & Instrumentation

摘　　要：针对机械手系统具有非线性时变、多变量、强耦合的特点,提出一种基于RBF神经网络逆系统的机械手解耦控制策略。首先证明了系统的可逆性,进一步通过神经网络在线逆辨识建立机械手的神经网络逆系统模型,并将辨识得到的逆模型作为控制器模型与机械手系统串联,构成伪线性复合系统,实现了将具有强耦合特性的多变量输入/输出机械手系统解耦成单个独立的伪线性对象。最后以两关节机械手为仿真对象进行了仿真,仿真结果验证了本方案的有效性和可行性。Based on the RBF neural network inverse system ,this paper presents a decoupling control strategy of manipulator, this strategy is in accordance with characteristics of manipulator system such as highly nonlinear, time-variant, multivariable and strong coupling. First the reversibility of system is testified,then the neural network inverse system of manipulator is established by neural network on-line identifiction, and the inverse model as the controller model and manipulator system in series, which forms a dynamic pseudo linear system. The multiple-input-multiple-output（MIMO） manipulator system with strong coupling is converted into isolated dynamic decoupling pseudo linear system. Finally, a simulation of the control scheme for a two-link robot has been made. The simulation results show the validity and feasibility of the scheme.

关 键 词：RBF神经网络 逆系统 机械手 最近邻聚类算法 解耦 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]