基于主成分分析-支持向量机算法的绝缘子污秽太赫兹检测研究  

Research on Terahertz Detection of Insulator Pollution Based on PCA-SVM Algorithm

作　　者：刘通 刘荣海[1] 李宗红 高青风 LIU Tong;LIU Ronghai;LI Zonghong;GAO Qingfeng(Electric Power Research Institute of Yunnan Power Grid Co.,Ltd.,Kunming,Yunnan 650217,China;North China Electric Power University,Beijing,102206,China)

机构地区：[1]云南电网有限责任公司电力科学研究院,云南昆明650217 [2]华北电力大学,北京102206

出　　处：《山西电力》2024年第3期6-11,共6页Shanxi Electric Power

基　　金：云南电网公司基金项目(YNKJXM20220050)。

摘　　要：为解决绝缘子污秽程度检测问题,提出了一种基于太赫兹时域光谱技术结合主成分分析-支持向量机模型的绝缘子污秽程度的非接触式检测方法。使用太赫兹时域光谱系统对不同污秽程度的人工污秽样本进行检测,获取样本的太赫兹光谱数据;提取样本的吸收系数用于主成分分析,建立支持向量机分类模型用于划分样本有无污秽、样本的污秽程度。结果表明,硅橡胶表面污秽程度在时域、频域、折射和吸收谱图上没有明显差异;基于吸收谱数据的主成分分析-支持向量机模型区分出有无污秽准确度为100%;基于吸收谱数据建立的支持向量机多分类模型对于测试集的识别准确率为92.5%。太赫兹时域光谱技术可应用于绝缘子污秽程度的检测,为绝缘子污秽程度的检测提供了新方法。In order to solve the pollution detection problem of insulators,a non-contact detection method for insulator pollution based on terahertz time-domain spectroscopy combined with principal component analysis-support vector machine(PCA-SVM)model is proposed in this paper.The terahertz time-domain spectral system is used to detect the artificial pollution samples of different pollution degrees and obtain the terahertz spectral data of the samples.What’s more,the absorption coefficients of extracted samples are used for principal component analysis and a support vector machine classification model is established to classify whether the samples are polluted or not and the pollution degree of the samples.The results show that there is no obvious difference in the pollution degree of silicone rubber surface in time domain,frequency domain,refraction and absorption spectrum.The accuracy of PCA-SVM model based on absorption spectrum data is 100%and the recognition accuracy of the SVM multi-classification model based on absorption spectrum data is 92.5%.Terahertz time-domain spectroscopy can be applied to the detection of the pollution degree of insulators,which provides a new method for the detection of pollution degree of insulators.

关 键 词：太赫兹 绝缘子 污秽程度 支持向量机 主成分分析 

分 类 号：TM855[电气工程—高电压与绝缘技术]