基于径向基函数神经网络的多关节机器人滑模控制器

Sliding Mode Control for Multi-joint Robot Based on RBF Neural Network

针对具有不确定性的多关节机器人系统,提出了一种径向基函数神经滑模控制方法;该控制方案采用全局滑模面,将神经网络的非线性映射能力与滑模控制的特点相结合,利用径向基神经网络自适应学习系统不确定性的未知上界,消弱了由滑模控制产生的抖动,同时保证了系统的鲁棒性;基于李亚普诺夫定理给出了系统稳定性的充分条件;仿真结果表明,该方法具有良好的轨迹跟踪和速度跟踪性能,提高了对于建模误差和不确定干扰等因素的鲁棒性。

A neural sliding mode controller is given for trajectory tracking control of multi--link robots with uncertain external disturb- ances and system model errors. This control scheme uses global sliding surface, integrates the characteristic of sliding mode control theory and the nonlinear mapping of neural network, Radial Basis Function neural network is applied to learn the unknown bounds of system uncer- tainties, reduce the chattering of sliding mode controller, at the same time the system has strong robustness. Based on the Lyapunov princi- ple, sufficient conditions for system stability are given. Simulation results verify that this method improves the performances of trajectory tracking and speed tracking, enhances the robustness to modeling error and external disturbances.