摘要
本文提出了一种直接控制的神经元模型，并用它构成了前馈神经网络，该网络在用ＢＰ学习算法学习时，提出了变学习步长和势态项系数的方法。并且该网络在特殊情况下，其输出可直接受输入神经元控制，把该神经元所构成的前馈网络应用于列车停车控制中。
A direct controlled neuron model is proposed and used for constructing feedforward neural network.And,an approach concerning the variation of learning step and the momentum rate coefficient is also presented when the network is trained by back-propagation algorism training.In addition,the outputs of this neural network can be controlled by input neuron under some special circumstances.This neuron-based feedforward neural network has been applied in the control system for train stopping.
出处
《西南交通大学学报》
EI
CSCD
北大核心
1995年第6期678-682,共5页
Journal of Southwest Jiaotong University
基金
国家八六三计划资助项目
关键词
神经元
神经网络
BP算法
机车控制
neuron
artificial neural network
back-propagation algorism
train control
