机械臂神经网络非奇异快速终端滑模控制

针对多自由度机械臂轨迹跟踪控制系统存在收敛速度慢、跟踪精度低的问题,提出了一种基于径向基神经网络(RBFNN)的非奇异快速终端滑模(NFTSM)自适应轨迹跟踪控制方法。首先,该方法采用非奇异快速终端滑模超曲面,切换控制项引入连续终端吸引子,使得系统能在有限的时间内收敛到平衡点。其次,采用RBFNN逼近系统未知非线性动力学,并结合逼近误差的自适应补偿机制,实现无模型控制。利用Lyapunov理论证明闭环系统的全局渐进稳定性和有限时间收敛性。最后,将该控制方法应用于Denso串联机械臂进行实验验证,并分析系统传输延时对实验结果的影响,提出解决方法。仿真和实验结果表明,该控制方法能有效地提高系统收敛速度和跟踪精度,增强对外部扰动的鲁棒性,削弱系统抖振。

Nonsingular Fast Terminal Sliding Mode Control of Robotic Manipulators Based on Neural Networks

A nonsingular fast terminal sliding mode adaptive controller based on RBF neural network was proposed for trajectory tracking control of multi degree of freedom manipulator with slow convergence speed and low tracking precision. Firstly, the nonsingular fast terminal sliding mode hypersurface was adopted in the control scheme and the continuous terminal attractor was introduced into the switch control, which made the system converge to the equilibrium point in a finite time. Secondly, the adaptive RBF neural network was used to approximate the unknown nonlinear dynamics of the system, the adaptive compensation mechanism of approximation error and adaptive law of weights of neural networks were designed to realize the model free control. The global asymptotic stability and finite time convergence of the closed-loop system were proved by Lyapunov theory. Finally, the control method was applied to Denso serial manipulator for experimental verification, the effect of transmission delay on the experimental results was analyzed and the solution was proposed. The simulation and experimental results demonstrated that the proposed control method can improve the convergence speed and the tracking accuracy of the system effectively, and enhance the robustness of the external disturbance. At the same time, it can weaken the chattering of the system and enhance the real-time control.