基于函数联接神经网络液压驱动机械手的自学习控制

Self-Learning Control of the Robot Arm Based on the Neural Network Connected with Functions

针对两自由度液压驱动机械手的轨迹跟踪控制问题 ,分析研究了神经网络自适应控制的特点 ,在此基础上提出了基于函数联接神经网络液压驱动机械手的自学习控制结构与控制算法·其控制结构着重智能知识的加强 ;控制算法以PD形学习规则为基础 ,运算结果不是直接参与控制 ,而是根据控制器作用于系统之后所产生的误差及其微分对控制器作出评价和修正·仿真和实验研究都取得了比较好的控制效果·

Toward the problem of the control of robot arm driven by hydraulic system, a new strategy dealing with the control of the robot arm was introduced, especially the problem of the control of tailing after the given orbit. This control method focus on enhancing the ability of artificial intelligent of the neural network. The learning rule is based on the PD control, but the result of the PD control is not used to control the object directly. The output of the PD controller is a training signal. The neural network gets the system error and the differential of the error system, and then correct and evaluate the weight of the neural network.