金属氧化物避雷器缺陷诊断的反距离加权改进KNN算法  

Inverse distance weighted improved KNN algorithm for defect diagnosis of MOA

作　　者：陈阳阳 舒胜文[1] 吴涵 王国彬 陈诚 CHEN Yangyang;SHU Shengwen;WU Han;WANG Guobin;CHEN Cheng(College of Electrical Engineering and Automation,Fuzhou University,Fuzhou,Fujian 350108,China;Electric Power Research Institute of State Grid Fujian Electric Power Co.,Ltd.,Fuzhou,Fujian 350007,China)

机构地区：[1]福州大学电气工程与自动化学院,福建福州350108 [2]国网福建省电力有限公司电力科学研究院,福建福州350007

出　　处：《福州大学学报（自然科学版）》2022年第5期635-641,共7页Journal of Fuzhou University(Natural Science Edition)

基　　金：福建省自然科学基金资助项目(2021J01635)。

摘　　要：针对现行金属氧化物避雷器(metal oxide arrester,MOA)在线监测系统中出现的因缺陷诊断规则不完善而导致大量漏报和误报事件,通过分析避雷器三相全电流和阻性电流、三相电压和阻性电流的Pearson相关系数,并考虑环境因素影响,提取环境温湿度、三相阻性电流和三相电压作为避雷器缺陷诊断的特征参数,提出一种基于反距离加权改进KNN(K-nearest neighbor)算法的避雷器缺陷诊断方法.通过实例验证,表明所提方法较其他方法具有更优的诊断正确率(97.28%)和泛化能力,为避雷器缺陷诊断提供了新思路.The defect diagnosis rules in the current metal oxide arrester(MOA)on-line monitoring system are still incomplete,resulting in a large number of false alarms and missing alarms.By analyzing the Pearson correlation coefficients of the three-phase full current and resistive current,three-phase voltage and resistive current of the arrester,and considering the influence of environmental factors,the ambient temperature and humidity,three-phase resistive current and three-phase voltage are extracted as the characteristic parameters of the arrester defect diagnosis.An arrester defect diagnosis method based on the inverse distance weighting improved K-nearest neighbor(KNN)algorithm is proposed,the case analysis is shown that the proposed method has better diagnosis accuracy(97.28%)and generalization ability than other methods,which provides a new idea for the diagnosis of arrester defects.

关 键 词：金属氧化物避雷器 改进KNN算法 在线监测 缺陷诊断 特征参数 

分 类 号：TM86[电气工程—高电压与绝缘技术]