Zhang Xueqin, Zhou Zhipeng, etc. of Southwest Jiaotong University： Study on Hyperspectral Detection Method of Contamination Grade of Insulators with Different Materials.

　　Reading Tip: This article is about 1700 words.

　　In order to meet the needs of intelligent development of UHV project in power system, a non-contact, efficient and fast detection method for contaminated insulators is proposed by using the advantages of hyperspectral, multi-band and high precision, which provides a new method for insulator surface contamination level detection. Based on the hyperspectral contamination detection technology, this paper realizes the contamination level detection of insulators with different materials under the same model by subsection direct standardization, which proves the great potential of hyperspectral technology in insulator contamination detection.

　　Research background

　　Insulators, as insulation devices for supporting conductors and transmission towers, conductors and internal structures, are widely used, and their materials mainly include ceramics, glass and composite insulators. Due to long-term exposure to air, their surface contamination may cause pollution flashover on the insulator surface. The traditional insulator contamination detection method has low efficiency and high misjudgment rate, and can not accurately detect insulator surface contamination. The hyperspectral insulator contamination detection technology has the advantages of multi-band, high precision and high efficiency.

　　The problems and significance of this paper.

　　At present, hyperspectral technology does not consider the influence of different materials in insulator pollution detection. The difference between different materials will be shown in hyperspectral spectral data, and it is impossible to detect the pollution level by the same model. Modeling the spectra measured by each material sample separately is costly, time-consuming and inefficient. Establishing a unified model to solve the detection and identification errors caused by different materials is the key to the application of hyperspectral technology in transmission line insulators.

　　The method and innovation of the paper

　　In this paper, hyperspectral technology is used to collect the data of pollution levels of insulating sheets made of different materials, and after relevant pretreatment, the main detection model of support vector machine (SVM) is established based on the experimental data of glass materials, and the main detection model is transmitted by piecewise direct standardization (PDS), so that the pollution levels of samples made of different materials under the same detection model can be detected.

　　1. Hyperspectral data processing and analysis.

　　The original spectral line data collected is the absolute reflection value of the sample to light. Environmental noise and dark current in the camera have influences on the data. There are many spectral line noises and interference information, so it is necessary to correct the collected data in black and white.

　　Fig. 1 Original map information

　　When the light is reflected on the surface of the material, some light will be scattered. Therefore, after black-and-white correction, scattering correction is needed to accurately represent data information. The main model of insulator pollution level detection based on support vector machine is established by using tag set data. The main flow chart is shown in the following figure.

　　Fig. 2 detection flow chart

　　2. Model establishment and result analysis

　　In this paper, support vector machine is selected to establish pollution level detection model, which is a commonly used generalized linear classifier with short training time and low complexity. A detection model of insulator contamination level based on support vector machine (SVM) is established by using the spectral line data of glass samples with different contamination levels and the spectral line data of detection.

　　Fig. 3 Average spectral line data of samples of different materials without contamination.

　　After the ceramic material detection set is corrected by PDS algorithm, the difference between the average spectral lines of the detection set of samples with different contamination levels of ceramic insulation sheets and the average spectral lines of the label set of samples with different contamination levels of glass insulation sheets is obviously reduced, and the corresponding spectral lines of the same contamination level are basically the same.

　　Fig. 4 The corrected average spectrogram of glass material label set and ceramic material detection set with different pollution levels.

　　PDS is used to correct the samples of silicone rubber material test set, and after the model transfer is realized, the corrected average spectral lines of different pollution levels are obtained, which are basically consistent with the spectral line data of glass material.

　　Fig. 5 The corrected average spectrogram of glass material label set and silicone rubber material detection set with different pollution levels.

　　conclusion

　　(1) Under the condition of the same pollution level of insulators made of different materials, the hyperspectral spectral lines are obviously different, which shows that the positions and changing trends of absorption peaks and reflection peaks are completely different; In the case of the same material with different pollution levels, the difference of spectral lines is mainly manifested in the amplitude. The higher the pollution level, the lower the spectral line amplitude.

　　(2) The SVM contamination level measurement model based on the label set of glass insulation samples has an accuracy of 98.3%. PDS algorithm can effectively reduce the difference between the spectral lines of ceramic and silicone rubber samples and glass samples.

　　(3) The detection method proposed in this paper can detect the contamination level of artificially contaminated insulators with different materials, with an accuracy rate of 83.3%, which provides a technical reference for the contamination level detection of insulators with different materials.

　　Author's introduction

　　Zhang xueqin

　　Associate Professor, Doctoral Supervisor, School of Electrical Engineering, Southwest Jiaotong University, distinguished professor, winner of Sichuan "Thousand Talents Program", candidate for academic and technical leaders in Sichuan Province, and member of "National Key Field Innovation Team" of Ministry of Science and Technology. Professor Zhang Xueqin devoted himself to the scientific research of external insulation discharge and key detection technologies in complex environment. He published more than 60 SCI/EI retrieval papers in top international academic journals, authorized more than 10 national invention patents, and won the first prize of the Science and Technology Progress Award of the Ministry of Education and the "Science and Technology Progress Award" of the State Grid Sichuan Electric Power Company.

　　The achievement of this work was published in the seventh issue of Journal of Electrotechnical Technology in 2023, and the title of the paper was "Study on Hyperspectral Detection Method of Contamination Grade of Insulators with Different Materials". This project is supported by the National Natural Science Foundation of China, the Outstanding Young Scientific Talents Project of Sichuan Province and the Science and Technology Project of State Grid Corporation of China.

　　Quote this article

　　Zhang Xueqin, Zhou Zhipeng, Guo Yujun, Yang Kun, Wu Guangning. Study on hyperspectral detection method of insulator pollution grade of different materials [J]. Journal of Electrotechnical Technology, 2023,38 (7): 1946-1955. Zhang Xueqin, Zhou Zhiping, 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.

　　Read the full text

　　china electrotechnical society

　　New media platform

　　Learn the official wechat

　　Journal of electrical technology

　　CES electricity

　　Official bilibili of the Society

　　CES TEMS

　　Today's headline

　　Learn popular science wechat

　　Tik Tok.

　　Journal of Electrical Technology: 010-63256949/6981; E-mail: dgjsxb@vip.126.com Electrical Technology: 010-63256943; E-mail: dianqijishu@126.com Journal of Electrical Machinery and Systems of China Electrotechnical Society (CES TEMS): Tel: 010-63256823; E-mail: cestems@126.com Editor: 010-63256994 Subscription: 010-63256817 Business Cooperation: 010-63256867/6838.