基于PCA-kmeans和图谱特征的绝缘子表面污秽程度评估方法  

Evaluation of Insulator Surface Pollution State Based on PCA-kmeans and Spectrum Features

作　　者：赵云龙 刘占双 李岩 潘超 贾明辉 任明 夏昌杰 ZHAO Yunlong;LIU Zhanshuang;LI Yan;PAN Chao;JIA Minghui;REN Ming;XIA Changjie(State Grid Qinghai Maintenance Company,Xining 810021,China;School of Electrical Engineering,Xi'an Jiaotong University,Xi'an 710049,China)

机构地区：[1]国网青海省电力公司检修公司,西宁810021 [2]西安交通大学电气工程学院,西安710049

出　　处：《电瓷避雷器》2023年第5期193-200,共8页Insulators and Surge Arresters

基　　金：国网青海省电力公司科技项目资助(编号:SGQHJX00YJJS2100361)。

摘　　要：绝缘子污秽闪络是威胁输电线路外绝缘防护的主要故障,为了寻求一种可靠的非接触式绝缘子污秽状态在线检测方法,本研究以高光谱成像技术为基础,通过人工污秽试验制备了5个污秽等级下不同类型的污秽绝缘子样品,随后通过高光谱成像平台分别获取了不同污秽状态下绝缘子样品的图谱特征并进行了分析,在此基础上结合主成分分析法和k-means空间聚类实现了绝缘子表面污秽等级的评估,并结合主成分计算结果和伪彩处理可以实现绝缘子表面污秽分布的可视化。结果表明该方法可以准确地实现污秽状态的分类和识别,识别准确度均在93%以上,其中复合绝缘子的污秽等级识别率最高到达了96.5%。本研究为绝缘子污秽状态的可视化在线监测提供了一种新的技术手段,同时为后续高光谱成像技术的现场污秽检测应用提供了研究基础。Insulator flashover is the main fault that threatens the external insulation protection of transmission lines.In order to seek a reliable non-contact online detection method of insulator pollution,based on hyperspectral imaging technology,the author prepared different types of insulator pollution samples under five pollution levels through artificial pollution test.Then set by highlights like platform respectively to obtain the graph characteristics of the insulator samples under different state of filth and analyzed,on the basis of the combination of principal component analysis(PCA)and k-means spatial clustering to realize the evaluation of the insulator surface pollution level,and the results combined with principal component and pseudo color processing can realize the visualization of the distribution of polluted insulator surface.The results show that this method can accurately classify and identify the pollution state,and the identification accuracy is above 93%,among which the highest identification rate of pollution grade of composite insulator is 96.5%.The research provides a new technical means for the visual online monitoring of insulator pollution,and also provides a research foundation for the subsequent application of hyperspectral imaging technology for on-site pollution detection.

关 键 词：污秽绝缘子 高光谱成像 污秽程度评估 可视化展示 

分 类 号：TM216[一般工业技术—材料科学与工程] TP391.41[电气工程—电工理论与新技术]