电工技术学报  2023, Vol. 38 Issue (7): 1946-1955    DOI: 10.19595/j.cnki.1000-6753.tces.211892
高电压与放电 |
不同材质绝缘子污秽等级高光谱检测方法研究
张血琴, 周志鹏, 郭裕钧, 杨坤, 吴广宁
西南交通大学电气工程学院 成都 611756
Detection Method of Contamination Grades of Insulators with Different Materials Based on Hyperspectral Technique
Zhang Xueqin, Zhou Zhipeng, Guo Yujun, Yang Kun, Wu Guangning
School of Electrical Engineering Southwest Jiaotong University Chengdu 611756 China
全文: PDF (8756 KB)   HTML
输出: BibTeX | EndNote (RIS)      
摘要 该文提出了适用于不同材质(陶瓷、玻璃和硅橡胶)绝缘子表面污秽等级的高光谱检测方法,采集不同材质和污秽等级样本的高光谱数据,经预处理后提取标签集及检测集谱线数据;采用玻璃材质样本标签集数据建立污秽等级检测模型;并运用分段直接标准化校正陶瓷、硅橡胶样本谱线数据,实现同一模型下不同材质样本的污秽等级检测。结果表明:不同材质样本同一污秽等级下,高光谱谱线吸收峰、反射峰位置及变化趋势有明显差异;同一材质不同污秽等级谱线差异主要为幅值。检测模型对玻璃、陶瓷和硅橡胶样本的污秽等级检测准确率分别为98.3%、95.0%和91.7%,并利用人工积污试验对模型进行了验证,污秽等级检测准确率为83.3%,证明了该模型可有效实现不同材质绝缘子表面污秽等级的高光谱检测。
服务
把本文推荐给朋友
加入我的书架
加入引用管理器
E-mail Alert
RSS
作者相关文章
张血琴
周志鹏
郭裕钧
杨坤
吴广宁
关键词 绝缘子污秽等级高光谱技术分段直接标准化支持向量机    
Abstract:The surface of naturally contaminated insulators is not completely covered by pollution and contains insulator material information. According to the detection principle of hyperspectral technology, the differences between porcelain, glass and silicone rubber insulator materials will be reflected in hyperspectral spectral line data, so it is hard to detect the contamination grades by the same model. Modeling the measured spectra of each material sample separately requires a large number of samples with known spectra and measured values, high modeling cost, long time and low efficiency. This paper realizes the rapid detection of the contamination grades of different material insulators through the piecewise direct standardization (PDS) model transfer method.
Firstly, test samples with different materials and different contamination grades were prepared according to relevant standards, and their hyperspectral data were obtained by the hyperspectral data acquisition platform, and their salt density and ash density values were measured using the equivalent salt deposit density. Then, the samples hyperspectral data after removing the interference were obtained through preprocessing operation. Finally, the main model of contamination grades detection of support vector machine is established based on the data of glass insulator, and the PDS algorithm is used to transfer the data of porcelain and silicone rubber material samples. The contamination grades detection method of different material insulation samples are obtained.
Through spectral line analysis and test verification, the following conclusions are drawn: (1) Under the same contamination grades, the hyperspectral lines of different material insulators have obvious differences, which are shown in the positions and changing trends of absorption peak and reflection peak. In the case of different contamination grades of the same material, the difference of spectral lines is mainly reflected in the amplitude. The higher the contamination grades, the lower the spectral line amplitude. (2) The support vector machine (SVM) contamination grades detection model was established by the data of glass insulation samples, and the detection accuracy of the contamination grades of glass samples was 98.3%. PDS algorithm can effectively reduce the difference between the spectral lines of porcelain, silicone rubber and glass samples. After the model transfer, the accuracy of the model in classifying the contamination grades of porcelain samples is increased from 25.0% to 95.0%, and the accuracy of the model in classifying the contamination grades of silicone rubber samples is increased from 25.0% to 91.7%. (3) The detection method proposed in this paper can detect the contamination grades of artificial contaminated insulators of different materials, and the accuracy is 83.3%.
Key wordsInsulator    contamination grade    hyperspectral technique    piecewise direct standardization    support vector machine   
收稿日期: 2021-11-18     
PACS: TM216  
基金资助:国家自然科学基金项目(51907168)、四川省杰出青年科技人才项目(2020JDJQ0039)和国家电网有限公司科技项目(521104190007)资助
通讯作者: 郭裕钧 男,1989年生,博士,副教授,研究方向为高电压与绝缘技术和输电线路防灾减灾技术。E-mail:yjguo@swjtu.edu.cn   
作者简介: 张血琴 女,1979年生,博士,副教授,博士生导师,研究方向为高电压绝缘与防护技术。E-mail:xq_zhang@home.swjtu.edu.cn
引用本文:   
张血琴, 周志鹏, 郭裕钧, 杨坤, 吴广宁. 不同材质绝缘子污秽等级高光谱检测方法研究[J]. 电工技术学报, 2023, 38(7): 1946-1955. Zhang Xueqin, Zhou Zhipeng, Guo Yujun, Yang Kun, Wu Guangning. Detection Method of Contamination Grades of Insulators with Different Materials Based on Hyperspectral Technique. Transactions of China Electrotechnical Society, 2023, 38(7): 1946-1955.
链接本文:  
https://dgjsxb.ces-transaction.com/CN/10.19595/j.cnki.1000-6753.tces.211892          https://dgjsxb.ces-transaction.com/CN/Y2023/V38/I7/1946