摘要
本文阐述了基于神经网络的继电保护系统的优越性；论证了由单层感知器网络或ＴＨ网络可以实现最小二乘算法，这两种网络都可以在极短的时间（数纳秒或几百纳秒）内完成全部运算；给出了电流继电器、圆特性以及四边型特性阻抗继电器的神经网络模型，并证明了三种模型都具有很强的自适应性。
The author expounds the suporiority of the protective system based on neuraI network and demonstrates that single-layer perceptron network or TH network can realize the least squares algorithm.Both networks may finish all the calculations in a extreme short time(several ns or several hundred ns).The paper presents the neural network models of the current relay,impedance relay of circle character and polygon character,and proves that these models have better adaptability.
出处
《西南交通大学学报》
EI
CSCD
北大核心
1995年第6期694-700,共7页
Journal of Southwest Jiaotong University
关键词
神经网络
继电保护
算法
neural network
protective relay
algorithm
