基于BP神经网络的输电线路隐患预放电识别研究  

Study on predischarge recognition of hidden dangers in transmission lines based on BP neural network

作　　者：杨旗 曾华荣 黄欢 马晓红 毛先胤 张露松 YANG Qi;ZENG Huarong;HUANG Huan;MA Xiaohong;MAO Xianyin;ZHANG Lusong(Electric Power Research Institute of Guizhou Power Grid Co.,Ltd.,Guiyang 550002 Guizhou,China;Key Laboratory of Ice Prevention&Disaster Reducing of China Southern Power Grid Co.,Ltd.,Guiyang 550002 Guizhou,China)

机构地区：[1]贵州电网有限责任公司电力科学研究院,贵州贵阳550002 [2]南方电网有限责任公司防冰减灾重点实验室,贵州贵阳550002

出　　处：《电力大数据》2020年第3期47-54,共8页Power Systems and Big Data

摘　　要：针对输电线路在长期运维过程中出现的异常,主要依靠人工定期巡线来排查,存在无法高效、准确的对隐患进行预判的局限,本文提出了基于BP神经网络方法对输电线路典型隐患预放电识别。首先在南方电网防冰减灾重点实验室梅花山基地搭建起输电线路放电试验平台,得到了输电线路在树障和污秽绝缘子两种典型隐患的预放电脉冲电流波形数据。继而绘制得到放电信号中放电量,放电次数,相位参数的三个二维统计图,在此基础上提取并构建得到放电特征参量数据库,再将特征量带入反向传播神经网络分类器中对线路隐患模型进行训练。最后对模型测试的结果表明,采用BP神经网络算法能够有效识别输电线路中典型的隐患,且识别准确率达到92%以上,进而为输电线路隐患识别提供了参考。Aiming at the abnormal state of transmission line in the long-term operation and maintenance process,the manual periodic line inspection method is mainly used for troubleshooting,and the hidden danger cannot be predicted efficiently and accurately. This paper presents a method for pre-discharge identification of typical hidden dangers in transmission lines based on BP neural network. Firstly,a typical test platform for transmission lines is established using key laboratory of ice prevention and disaster reducing of China southern power grid in Plum Blossom Hill. the pre-discharge pulse current waveform data of two typical hidden dangers of transmission lines in tree barriers and dirty insulators were obtained. Then draw three two-dimensional statistical graphs of discharge amount,discharge times,and phase parameters in the discharge signal. On this basis,the discharge characteristic parameter database is extracted and constructed,and then the characteristic quantity is brought into the back propagation neural network classifier to train the hidden danger model of the line. Finally,the model test results show that the BP neural network algorithm can effectively identify the typical hidden dangers in transmission lines,and the recognition accuracy is more than 92%,which provides a reference for the hidden dangers identification of transmission lines.

关 键 词：神经网络 输电线路 隐患 模拟试验 放电 识别 

分 类 号：TM72[电气工程—电力系统及自动化]