论文部分内容阅读
提出了一种基于观测器的机械手神经网络自适应轨迹跟随控制器设计方法，这里机械手的动力学非线性假设是未知的，并且假设机械手仅有关节角位置测量．文中采用一个线性观测器重构机械手的关节角速度，用神经网络逼近修正的机械手动力学非线性，改进系统的跟随性能．基于观测器的神经网络自适应控制器能够保证机械手角跟随误差和观测误差的一致终结有界性以及神经网络权值的有界性，最后给出了机械手神经网络自适应控制器－观测器设计的主要理论结果，并通过数字仿真验证了所提方法的性能．
An observer-based design method of manipulator neural network adaptive trajectory following controller is proposed. Here, the assumption of nonlinear dynamics of manipulator is unknown, and it is assumed that manipulator only has joint angle measurement. In this paper, a linear observer is used to reconstruct the joint angular velocity of the manipulator. The neural network is used to approximate the corrected nonlinearity of the manipulator dynamics to improve the following performance of the system. The observer-based neural network adaptive controller can guarantee the uniform end-followability of robot angle error and observational error and the boundedness of the weights of neural network. Finally, an adaptive controller-observer design of manipulator neural network The main theoretical results are given and the performance of the proposed method is verified by digital simulation.