缪金,秦军,陈峻宇,吴俊锋,任明,刘润宇
MIAO Jin,QIN Jun,CHEN Junyu,WU Junfeng,REN Ming,LIU Runyu

• 1:国网江苏省电力有限公司无锡供电分公司
• 2:西安交通大学电力设备电气绝缘国家重点实验室

摘要(Abstract)：

复合绝缘子表面的积污成分是影响污秽闪络电压的重要因素。为了探索更为高效且实用化的典型污秽成分识别方法，提出了一种基于高光谱成像的污秽成分识别方法。首先，通过标准人工染污方式制备了7种典型污秽成分的单一或混合积污硅橡胶样品，并在模拟日光条件下获得这些积污样品的高光谱二维像元数据。接着，通过在二维像元维度引入KPCA（核主成分分析），有效提取典型污秽成分的像元数据特征，并在光谱维度上基于包络线消除和KPCA建立三维特征子空间。最后，引入KSVM（核支持向量机）算法实现了对7种污秽成分像元的识别。结果表明，该方法对硅橡胶表面积污成分的识别准确率达到了93.75%，而对于混合污秽成分的识别准确率高于80%。研究成果为复合绝缘子污秽成分的现场快速分析和闪络特性评估提供了一种新手段，同时也为带电巡线和实验室污秽度测定提供了参考。
The surface contamination components of composite insulators significantly influence the pollution flashover voltage. To develop a more efficient and practical method for identifying typical contamination components, this paper proposes a hyperspectral imaging-based identification method. Firstly, silicone rubber samples with single or mixed contamination of seven typical pollutants are prepared using standard artificial contamination methods, and hyperspectral 2D pixel data of these samples are acquired under simulated sunlight conditions. Next, kernelized principal component analysis(KPCA) is introduced at the 2D pixel level to effectively extract pixel data features of the typical contamination components. A 3D feature subspace is then established in the spectral dimension using envelope elimination and KPCA. Finally, the kernel support vector machine(KSVM) algorithm is employed to identify the pixel features of the seven contamination components. Results show that the proposed method achieve an identification accuracy of 93.75% for contamination components on the silicone rubber surface and over 80% for mixed contaminants. This study provides a novel approach for rapid on-site analysis of contamination components and flashover characteristics of composite insulators, as well as a reference for live-line inspections and pollution degree measurements in laboratories.

关键词(KeyWords)： 绝缘硅橡胶;绝缘积污;反射光谱图像;污秽度;光谱特征提取
insulating silicone rubber;insulation contamination;reflective spectral imaging;contamination degree;spectral feature extraction

Abstract:

Keywords:

基金项目(Foundation): 国网江苏省电力有限公司孵化项目（JF2024002）

作者(Author): 缪金,秦军,陈峻宇,吴俊锋,任明,刘润宇
MIAO Jin,QIN Jun,CHEN Junyu,WU Junfeng,REN Ming,LIU Runyu

DOI: 10.19585/j.zjdl.202501013

参考文献(References)：

• [1]张柏林，郁娇山，黄万龙，等.甘肃地区高比例新能源大外送电网面临的挑战及思考[J].电网与清洁能源，2020,36(4):81-89.ZHANG Bolin,YU Jiaoshan,HUANG Wanlong,et al.Challenges and reflections on the power grid of high proportion of new energy in Gansu Province[J].Power System and Clean Energy,2020,36(4):81-89.
• [2]常波，张启哲，王胜辉，等.动车组车顶隔离开关用支柱绝缘子污秽状态下的电场分布特性研究[J].电测与仪表，2024,61(4):81-85.CHANG Bo,ZHANG Qizhe,WANG Shenghui,et al.Research on the electric field distribution characteristics under polluted conditionsof EMU roof isolation switch post insulators[J].Electrical Measurement&Instrumentation,2024,61(4):81-85.
• [3]王亚平，李智辉，阿思亚，等.高电场下直流绝缘子表面积污特性研究[J].山东电力技术，2024,51(8):78-84.WANG Yaping,LI Zhihui,A Siya,et al.Study on the contamination characteristics of DC insulators under high electric field[J].Shandong Electric Power,2024,51(8):78-84.
• [4]刘子英，肖建华，邓芳明.基于可见光图像识别的绝缘子污秽等级判别[J].传感器与微系统，2019,38(12):136-139.LIU Ziying,XIAO Jianhua,DENG Fangming.Identification of insulator contamination grade based on visible light image recognition[J].Transducer and Microsystem Technologies,2019,38(12):136-139.
• [5]张国治，田晗绿，杨帅，等.温度对GIS内部固体绝缘缺陷间歇性放电特性影响研究[J].电力工程技术，2024,43(3):183-191.ZHANG Guozhi,TIAN Hanlyu,YANG Shuai,et al.The impact of temperature on intermittent discharge characteristics of solid insulation defects inside GIS[J]. Electric Power Engineering Technology,2024,43(3):183-191.
• [6]李成学，吕邦欢.基于带电颗粒动态积污的绝缘子电场分布研究[J].高压电器，2024,60(5):147-155.LI Chengxue,LYU Banghuan. Research on electric field distribution of insulators based on dynamic pollution accumulation of charged particles[J].High Voltage Apparatus,2024,60(5):147-155.
• [7]孟晓波，曹针洪，杨旭洋，等.基于碰撞电离理论的污秽及环境因素对导线起晕电压的影响分析[J].智慧电力，2023,51(8):104-110.MENG Xiaobo,CAO Zhenhong,YANG Xuyang,et al.Influence of dirt&environmental factors upon corona initiation voltage based on collision ionization theory[J].Smart Power,2023,51(8):104-110.
• [8]孙磊，吴文海，柯坚，等.基于改进SVDD的绝缘子污秽检测方法[J].传感器与微系统，2019,38(9):57-59.SUN Lei,WU Wenhai,KE Jian,et al.Insulator pollution detection method based on improved SVDD[J]. Transducer and Microsystem Technologies,2019,38(9):57-59.
• [9]赵甦，宗飞.大气中污秽混合物对输电线路绝缘子的影响[J].科技风，2019,27:117.ZHAO Su,ZONG Fei. Effect of atmospheric pollution mixtures on transmission line insulators[J]. Technology Wind,2019,27:117.
• [10]蒋兴良，杨忠毅，韩兴波，等.硅橡胶复合绝缘子在不同可溶污秽成分下的闪络特性研究[J].中国电机工程学报，2018,38(1):320-329.JIANG Xingliang,YANG Zhongyi,HAN Xingbo,et al.Research on flashover performance of silicone rubber insulators polluted by different soluble pollution constituents[J].Proceedings of the CSEE,2018,38(1):320-329.
• [11]王乃啸，高海翔，王希林，等.基于BP神经网络的绝缘子污秽成分LIBS在线检测技术[J].广东电力，2020,33(9):49-57.WANG Naixiao,GAO Haixiang,WANG Xilin,et al.LIBS online detection technique for insulator contamination based on BP neural network[J].Guangdong Electric Power,2020,33(9):49-57.
• [12]李恒真，叶晓君，刘刚，等.广州地区输电线路沿线绝缘子自然污秽化学成分的来源分析[J].高电压技术，2011,37(8):1937-1943.LI Hengzhen,YE Xiaojun,LIU Gang,et al.Source analysis on the chemical composition of natural contamination on the line insulator in Guangzhou area[J].High Voltage Engineering,2011,37(8):1937-1943.
• [13]刘嵘，张燕，段玉兵，等.山东电网绝缘子表面污秽成分特点分析[J].电瓷避雷器，2017(4):209-214.LIU Rong,ZHANG Yan,DUAN Yubing,et al.Analysis on the characteristics of surface contamination composition of insulators in Shandong power grid[J]. Insulators and Surge Arresters,2017(4):209-214.
• [14]任绍梅，杨冬亮.高压绝缘子自然污秽中阴离子组成的分析[J].北京石油化工学院学报，2009,17(2):11-14.REN Shaomei,YANG Dongliang.The analysis of the anions in the deposits of natural polluted insulators[J].Journal of Beijing Institute of Petro-Chemical Technology,2009,17(2):11-14.
• [15]王永浩.复合绝缘子运行状态在线非接触检测技术研究[D].武汉：湖北工业大学，2016.WANG Yonghao.The research of operating status of the composite insulators based on the online non-contact detection technique[D].Wuhan:Hubei University of Technology,2016.
• [16] XIA C J,REN M,WANG S Y,et al.Pollution degree detection of insulators based on hyperspectral imaging technology[C]//2019 IEEE 20th International Conference on Dielectric Liquids(ICDL).June 23-27,2019.Roma,Italy:IEEE,2019:1-4.
• [17]庆雨豪.基于深度学习的高光谱图像分类算法研究[D].广州：中北大学，2023.QING Yuhao. Research on a deep learning-based algorithm for hyperspectral image classification[D]. Taiyuan.North University of China,2023.
• [18]吴静珠，吴胜男，刘翠玲，等.近红外和高光谱技术用于小麦籽粒蛋白含量预测探索[J].传感器与微系统，2013,32(2):60-62.WU Jingzhu,WU Shengnan,LIU Cuiling,et al.Explorations of wheat grain protein content prediction using NIR and hyperspectrum technology[J].Transducer and Microsystem Technologies,2013,32(2):60-62.
• [19]陈永聪.照明光源光谱检测研究[J].光源与照明，2022(6):76-78.CHEN Yongcong.Research on spectral detection of illumination light sources[J]. Light Source and Lighting,2022(6):76-78.
• [20] SHAWETAYLOR J.Kernel methods for pattern analysis[M].Beijing:China Machine Press,2005.
• [21]吴一全，吴超.结合NSCT和KPCA的高光谱遥感图像去噪[J].遥感学报，2012,16(3):533-544.WU Yiquan,WU Chao.Denoising of hyperspectral remote sensing images using NSCT and KPCA[J].Journal of Remote Sensing,2012,16(3):533-544.
• [22]刘方园，王水花，张煜东.支持向量机模型与应用综述[J].计算机系统应用，2018,27(4):1-9.LIU Fangyuan,WANG Shuihua,ZHANG Yudong,et al.Overview on models and applications of support vector machine[J].Computer Systems&Applications,2018,27(4):1-9.