基于RBF神经网络的输电线路无通道保护  

Noncommunication Protection for Transmission Lines Based on RBF Neural Network

作　　者：陈少华[1] 余耀权[1] 郑帅[1] 叶杰宏[1] 

机构地区：[1]广东工业大学自动化学院,广东广州510090

出　　处：《电力科学与工程》2004年第2期9-11,48,共4页Electric Power Science and Engineering

基　　金：广东省教育厅自然科学基金资助项目(Z02033)

摘　　要：在故障分析的基础上,提出了一种应用RBF神经网络实现输电线全线路无通道快速保护的原理,由85%范围内的速动保护和85%范围外的相继动作保护2部分构成。该网络采用3层神经网络模型,由4部分构成:方向判别子网络ANN1,故障类型及相别判断子网络ANN2,瞬时出口动作子网络ANN3及相邻断路器跳闸检测子网络ANN4。通过对各种故障类型的EMTP和Matlab的仿真测试,证明了该保护的可行性。On the basis of fault analysis, this paper presents the theory of using artificial neural network to implement the noncommunication fast protection in transmission lines. It consists of the instantaneous protection at the protective range of 85% and the sequence tripping protection beyond the protective range of 85%. The network, a threelayer RBF model, consists of ANN1 determining fault directions, ANN2 identifying fault types and fault phase, ANN3 tripping breaker instantaneously and ANN4 detecting trip situation of an adjacent breaker. Simulation results of all types of faults verified the model.

关 键 词：电流保护 无通道保护 RBF神经网络 相继动作 

分 类 号：TM71[电气工程—电力系统及自动化]