针对动力学模型中带有参数未知特性的机械臂位置和速度跟踪问题,提出一种新型的基于神经网络和全阶滑模的控制策略.该策略考虑了执行器的动力学特性,然后基于跟踪误差,建立了全阶滑模面,并利用径向基网络对模型未知特性进行逼近,进而设计鲁棒自适应控制律使得系统状态到达滑模面并沿滑模面收敛到平衡点.理论分析证明了所设计的控制策略可在克服抖振问题的同时保证闭环系统的渐近稳定性.二连杆机械臂系统的数值仿真结果验证了所提出方法的有效性.

For the position and velocity tracking problem of robotic manipulators whose dynamic models have unknown parameters, a novel full-order sliding mode control strategy is proposed by using the neural network. Based on the tracking errors of the robotic manipulators including actuator dynamics, a full-order sliding mode is established and the radial basis function network is utilized to approximate the unknown terms of the system model. A robust adaptive controller is designed to drive the system states to reach the sliding mode and then converge to the equilibrium along the sliding mode. Theoretical analysis demonstrates that the proposed control strategy can overcome the chattering problem and guarantee the asymptotic stability of the closed-loop system. Simulation results of a two-link robotic manipulator are presented to validate the proposed method.