基于神经网络的非驱动关节机械臂位置控制研究  

Underactuated manipulator position control by using neural network

作　　者：朱达书[1] 朱齐丹[1] 邢卓异[1] 

机构地区：[1]哈尔滨工程大学自动化学院,黑龙江哈尔滨150001

出　　处：《哈尔滨工程大学学报》2005年第3期307-310,共4页Journal of Harbin Engineering University

摘　　要：提出了一种对具有非驱动关节机器人基于神经网络的位置开环控制方法.首先在驱动臂施加振动扰动,使非驱动臂达到稳定平衡点,再通过神经网络学习驱动臂与非驱动臂的运动关系,利用该关系实现了非驱动臂的任意位置开环控制.在此基础上,设计开发了具有非驱动关节的平面2自由度机械手,并且将该方法实际应用于该机械手模型进行了控制实验,试验结果表明,该方法是可行的,实现了非驱动臂的任意位置控制.The position control of underactuated joint based on neural network is investigated. It makes the underactuated joint to stable position by putting continuous disturb on actuated joint. The relationship between the underactuated joint and actuated joint was studied to realize the open-loop arbitrary position control by neural network. An experiment set was designed for underactuated robot with 2-freedom-degree in the plane. Experimental results indicate that it is feasible to realize the open-loop random position control.

关 键 词：机器人 非驱动关节 神经网络 PID控制 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置]