基于反演设计的机械臂非奇异终端神经滑模控制

Nonsingular Terminal Neural Network Sliding Mode Control for Manipulator Joint Based on Backstepping

针对具有建模误差和不确定干扰的多关节机械臂的轨迹跟踪问题,设计反演非奇异终端神经滑模控制。该方案是采用能有限时间收敛的非奇异终端滑模面,根据滑模控制原理和反演方法设计反演滑模控制器;对于反演滑模控制系统中由于建模误差和不确定干扰造成的不确定因素的上界,设计径向基(Radial basis function,RBF)神经网络自适应律,在线估计不确定因素的上界;利用李亚普诺夫定理证明了系统的稳定性。仿真结果表明,该方法具有良好的轨迹跟踪性能,提高对于建模误差和不确定干扰等因素的鲁棒性,削弱了抖动。

A new method of nonsingular terminal neural network sliding control based on backstepping for tracking control of multi-link robot manipulators with modeling error and external disturbances is introduced. Nonsingular terminal sliding mode surface is used which has property of finite-time convergence. According to sliding mode control theory, backstepping is used to design nonsingular terminal sliding mode controller. To confn＇m the upper bound of uncertainties in sliding-mode control system, a proper radial basis function neural networks controller is designed to estimate uncertain upper boundary on line. The system stability is proved by Lyapunov principle simulation results verify that this method improves the performances of trajectory tracking, enhances the robustness to modeling error and external disturbances and reduces chattering.