摘要
针对可重构机械臂动力学中存在的模型参数摄动和外界扰动,本文阐述了一种基于速度观测模型的模糊RBF神经网络补偿控制算法.利用Lyapunov函数给出了网络的权值、隶属度函数中心和宽度倒数的在线更新律,并证明了所提出的观测模型及其补偿控制算法的最终一致有界性.最后以RRP(revolute-revolute-prismatic)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问题的有效性,同时与基于速度观测模型的RBF神经网络补偿控制进行了仿真对比及分析,给出了神经网络和模糊神经网络在可重构机械臂轨迹控制应用中各自的优缺点.
Parameter uncertainty and noise disturbance are unavoidable in the reconfigurable manipulator systems. To deal with this problem, we propose a velocity-observer-based neurofuzzy compensating control scheme. The update laws on weight, center and width-reciprocal of the membership function in the radial-basis-function-based (RBF) neurofuzzy are given by Lyapunov theorem. The proposed algorithm is proved to be ultimately uniformly bounded (UUB). The controller for a RRP (revolute- revolute- prismatic) reconfigurable manipulator is simulated and discussed. Simulation results show that the proposed algorithm is effective and satisfactory in tracking performance. Finally, simulation comparison between the velocity-observer-based RBF and the proposed neurofuzzy compensator is conducted and analyzed. Advantages and disadvantages for them are given.
出处
《控制理论与应用》
EI
CAS
CSCD
北大核心
2008年第5期891-897,共7页
Control Theory & Applications
基金
国家自然科学基金资助项目(60674091).