可重构机械臂鲁棒模糊神经补偿控制仿真研究  

Simulation on Robust Neurofuzzy Compensator for Reconfigurable Manipulator Motion Control

作　　者：李英[1] 朱明超[1] 李元春[1] 

机构地区：[1]吉林大学通信工程学院,长春130022

出　　处：《系统仿真学报》2007年第22期5169-5174,共6页Journal of System Simulation

基　　金：吉林省科技计划发展项目(20040532)。

摘　　要：基于精确模型计算力矩控制,考虑了可重构机械臂的动力学系统中存在的大量不确定性,提出了鲁棒模糊神经智能控制算法辨识补偿结构和非结构不确定性,给出了模糊神经网络的权值、隶属度函数参数以及综合误差的分立量估计Lyapunov稳定性在线调节律。通过Lyapunov稳定性理论证明了提出的RNF补偿算法的最终一致有界性,以RR(Revolute-revolute)和RRP(Revolute-revolute-prismatic)两种构形的可重构机械臂为例,通过仿真研究了算法对轨迹跟踪的有效性。Computing torque control （CTC） based robust neurofuzzy （RNF） control scheme was developed to identify and compensate structured and unstructured uncertainties, which were impossible avoided for reconfigurable manipulator. The update laws on weights, centers and widths of membership function in neurofuzzy network, and part of comprehensive error were given by Lyapunov stability theorem, Importantly, robust neurofuzzy algorithm proved ultimately uniform bounded （UUB） through Lyapunov stability theory. Finally, the controller for a RR and a RRP reconfigurable manipulator were designed and simulated due to different uncertainty resulted by different configuration; simulation results were discussed and show the proposed algorithm effective and satisfactory on tracking performance.

关 键 词：计算力矩控制 鲁棒模糊神经补偿 可重构机械臂 仿真 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]