人工神经网络在光学电压互感器故障诊断中的应用  

Application of Artificial Neural Network to Fault Diagnosis for Optical Voltage Transformer

作　　者：蒋愈勇 王军龙[2] 李俊一[2] 于文鹏[2] 陈波 李文伟 JIANG Yuyong;WANG Junlong;LI Junyi;YU Wenpeng;CHEN Bo;LI Wenwei(Smart Grid Technology Research Center,Electric Power Research Institute Company Limited,China Southern Power Grid,Guangzhou 510663,China;Beijing Aerospace Control Instrument Research Institute Beijing 100854,China;Qinzhou Power Supply Bureau,Guangxi Power Grid Company Limited,Qinzhou 535000,China)

机构地区：[1]南方电网科学研究院有限责任公司智能电网研究所,广州510663 [2]北京航天控制仪器研究所,北京100854 [3]广西电网有限责任公司钦州供电局,钦州535000

出　　处：《电力系统及其自动化学报》2018年第6期134-139,共6页Proceedings of the CSU-EPSA

摘　　要：随着光学电压互感器在电网的广泛使用,光学电压互感器的故障诊断也成为迫切需要解决的问题。本文提取光学电压互感器故障模式的形状特征、时域特征、频域特征和时频联合特征构成故障特征向量,之后,将故障模式特征向量作为输入对反向传播BP(back propagation)神经网络进行训练,从而实现对光学电压互感器的故障诊断。基于Matlab仿真实验获取数据验证了方法的可靠性和准确性。验证结果表明,本文所提出的基于BP神经网络的故障诊断方法可靠、准确,诊断正确率在90%以上。With the widespread applications of optical voltage transformer in power system,the fault diagnosis for optical voltage transformer has become an urgent problem. In this paper,the shape,time-domain,frequency-domain,andtime-frequency joint characteristics of the fault mode were extracted to form a fault feature vector at first. Then,the faultfeature vector was used as input to train the back propagation（BP）neural network,thus realizing the fault diagnosis foroptical voltage transformer. The reliability and accuracy of this method were verified based on sampling data generatedin simulation experiments using Matlab,showing that the proposed BP neural network based fault diagnosis method was reliable and accurate,and the accuracy rate of diagnosis was above 90%.

关 键 词：光学电压互感器 影响因素 特征参数 故障模式 人工神经网络 反向传播 

分 类 号：TM451[电气工程—电器]