刚性机械臂系统的抗饱和神经网络动态面控制  

Anti-windup Neural Network Dynamic Surface Control for Rigid Robot Manipulator

作　　者：姚月琴[1] 王影星[1] 张磊[2] YAO Yueqin;WANG Yingxing;ZHANG Lei(Yancheng Vocational Institute of Industry Technology, YanchengJiangsu 224005 , China;Xidian University, Xi'an Shanxi 710071, China)

机构地区：[1]盐城工业职业技术学院,江苏盐城224005 [2]西安电子科技大学,陕西西安710071

出　　处：《机床与液压》2019年第3期27-31,共5页Machine Tool & Hydraulics

基　　金：江苏省自然科学基金资助项目(BK20150039)

摘　　要：针对机械臂系统轨迹跟踪控制中存在外界干扰及控制输入受限问题,提出一种抗饱和神经网络动态面控制方法。其中模型中不确定项和外界干扰由径向基函数神经网络补偿,控制输入受限部分由辅助的抗饱和函数解决,完整的系统控制律由动态面控制方法获得。该控制算法解决了反步法中可能存在的"微分爆炸"现象,避免了滑模控制中存在的"抖振"现象以及自适应控制的鲁棒性受限现象。最后,设计相应的Lyapunov函数证明整个闭环系统的半全局渐进稳定性,仿真结果证实了所设计控制算法的有效性。Considering the external disturbances and the control input saturation constraint problems existing in the trajectory tracking control of the robot manipulators, an anti-windup neural network dynamic surface control algorithm was proposed. Whereas the terms of the model uncertainties and the external disturbances could be compensated by the radial basis function neural networks, and the saturation constraint problem could be solved by bringing into an auxiliary anti-windup function, the integral control law was obtained by the dynamic surface control law. The designed control law can solve the phenomenon of "differential explosion" existing in the traditional back-stepping control, avoiding the chattering phenomenon appearing in the sliding mode control, and the robustness constraint existing in traditional adaptive control. Finally, relevant Lyapunov function was designed to validate the semi-globally gradually stability of the closed loop system, and the simulation results further verified the credibility of the proposed controller.

关 键 词：机械臂 抗饱和 神经网络 动态面控制 

分 类 号：TP241.2[自动化与计算机技术—检测技术与自动化装置]