基于神经网络的机器人迭代学习控制

针对机器人动力学模型的不确定性和负载扰动，提出了一种采用神经网络的机器人迭代学习控制方法。该方法将反馈控制和神经网络学习控制相结合，反馈控制沿时间轴方向使关节运动跟踪期望轨迹，神经网络学习控制沿迭代轴方向使关节运动逼近期望轨迹。文中还给出了基于ＢＰ神经网络的学习控制算法。仿真结果表明，该方法能克服机器人动力学模型的不确定性和负载扰动，具有良好的鲁棒性和控制性能。

Iterative Learning Control for Robot Manipulators Using Neural Networks

Aiming at dynamic model uncertainties and load disturbances of robot manipulators, an iterative learning control scheme using neural networks is presented. The scheme combines a feedback control with a neural networks based learning control, in which the feedback control makes the movement of joints of robot manipulators track the desired trajectory along the direction of time axis and the learning control makes the movement of joints of robot manipulators approach the desired trajectory along the direction of iterative axis. The BP neural network based learning control algorithm is also given . The simulation results show that the presented scheme can overcome the model uncertainties and load disturbances and has a strong robustness and good control performance.