Abstract: In order to improve intelligence of automatic mechanisms and promote transformation from novel technology to productive force, this paper introduces the construction and application of the manipulator teaching system based on the four-layer perceptron neural network. It illustrates the design of four-layer perceptron neural network, the principles of supervised learning in this network, and a method of synthetic back propagation for the network with not differential link.

Key words: teaching system, four-layer perceptron neural network, method of synthetic back propagation, manipulator

摘要： 为了提升自动化机械智能水平，推动高新技术向生产力的转化，该文介绍了基于四层感知器神经网络的机械手示教系统的结构和应用实例，说明了四层感知器神经网络的设计、有导师学习的工作原理，以及针对含有不可微函数环节的神经网络所采用的综合反向传 播法。

关键词: 示教系统, 四层感知器神经网络, 综合反向传播法, 机械手

CLC Number: 

• TH165