摘要
CMAC神经网络已被广泛用于机械手的控制 ,但这种网络的连接权矩阵通常由学习得到 ,有时因学习时间长而影响控制的适时性。本文提出一种连接权矩阵的计算方法 ,不用学习可直接计算求得连接权矩阵 ,其均方根误差近似为零 ,从而实现了用于机械手控制的机械手 CMAC神经网络的记忆。保证了机械手控制的适时性。
CMAC neural network has been used in the control of robot hands. Its link weight matrix is usually obtained by a time-consuming learning process, which will certainly degrade the control effect. In this paper a method for computing link matrix is given. With this method, the learning process is eliminated, and the mean square error is almost zero, thus adding memory to CMAC neural network, and meeting the requirement of real-time control of robot hands.
出处
《机械科学与技术》
CSCD
北大核心
2001年第1期29-30,32,共3页
Mechanical Science and Technology for Aerospace Engineering
基金
863-51-945-011资助项目
