漂浮基空间机械臂姿态、关节协调运动的模糊小波神经网络控制  

Tracking Control for Coordinated Motion of Space Manipulator in Joint Space Based on Fuzzy Wavelet Neural Network

作　　者：黄登峰[1] 陈力[1] 

机构地区：[1]福州大学机械工程及自动化学院,福州350108

出　　处：《空间科学学报》2011年第6期821-828,共8页Chinese Journal of Space Science

基　　金：国家自然科学基金项目(10672040);福建省自然科学基金项目(2010J01003)共同资助

摘　　要：讨论了本体姿态受控、位置不受控制的漂浮基空间机械臂系统协调运动的动力学控制问题.根据系统位置几何关系、动量守恒关系和第二类拉格朗日方程,建立漂浮基空间机械臂系统的动力学方程.在此基础上,针对系统参数未知的情况,设计了一种采用小波基函数作为模糊隶属度函数的模糊神经网络控制器,以控制空间机械臂的本体姿态和机械臂两关节铰协调地完成各自在关节空间的期望运动.其特点是不要求系统动力学方程关于惯性参数呈线性函数关系,甚至不需要知道系统参数;而且网络权值是采用反向传播算法根据误差进行在线学习,使模糊神经网络获得更强的自学习和自适应能力,同时也节省了离线学习的时间.系统数值仿真的结果证实上述控制方案是行之有效的.The trajectory tracking control for coordinated motion of free-floating space manipulator system with an attitude controlled base is discussed. According to the geometrical relationship and law of conservation of momentum, the dynamic equations of space manipulator system are constituted through Lagrange equation of the second kind. Based on the above, a fuzzy wavelet neural network scheme is designed to control the base attitude and the joint angle of manipulator to track desired trajectories synchronously in joint space on condition that system parameters are unknown. The control scheme needs neither linearly parameterize the dynamic equations of the system, nor know any system parameters. Further more, its network weights are learning online based on errors by back propagation algorithm so the self-learning and self-adapting ability of fuzzy neural network is enhanced and meantime the offiine training time is saved. Numerical simulations are presented to verify the feasibility and effectiveness of the proposed control scheme.

关 键 词：漂浮基空间机械臂系统 模糊小波神经网络 协调运动 关节空间 

分 类 号：TP241[自动化与计算机技术—检测技术与自动化装置] TP273.4[自动化与计算机技术—控制科学与工程]