基于PCA和聚类的断路器分合闸线圈电流研究  

Research of the Circuit Breakers′Switching Coil Currents Based on PCA and Clustering Algorithm

作　　者：李春锋 孔海洋[2] 王璇[2] 方福歆 关伟民[2] 

机构地区：[1]中国平煤神马集团公司电务厂,河南平顶山467000 [2]武汉大学电气工程学院,武汉430072

出　　处：《电力与能源》2016年第1期32-36,共5页Power & Energy

摘　　要：断路器的机械特性故障可以通过分合闸线圈电流反应。通过对分合闸线圈的电流分析提取出7个时间和电流特征量,对所有特征量进行主分量分析(PCA),在保留原特征量信息的基础上,实现了多维数据的降维可视化。对降维后的数据通过K-means聚类分析,经过大量样本的学习,即可直观地区分断路器的正常与各种故障状态。该研究方法为高压断路器的实时在线故障诊断提供了一种新途径,与人工识别线圈电流波形判断故障相比,利用机器识别具有更高的可靠性和工作效率。The mechanical failure of circuit breakers can be responded by switching coil current. This paper ex- tracted seven time and current characteristics through the analysis of switching coil current, and analyzed all characteristics using Principal Component Analysis （PCA）. This method can not only retain the original infor mation of the characteristic~ but also implement the dimension reduction and visualization of multidimensional data. Through K-means clustering analysis of the data after dimension reduction, and after a large sample of study, we can distinguish between normal and all kinds of fault state of circuit breakers intuitively. The re- search method provides a new way for real-time online fault diagnosis of HV circuit breakers. Compared with judging fault using the artificial identification of the switching coil current waveform, the machine identification will have higher reliability and working efficiency.

关 键 词：高压断路器 线圈电流 PCA 聚类分析 

分 类 号：TM561[电气工程—电器]