基于高光谱与红外技术融合的绝缘子污秽程度检测方法  

Method of insulator pollution degree detection method based on fusion of hyperspectral and infrared technology

作　　者：刘益岑 杨琳 王杨宁 殷成凤 郭裕钧 张血琴 LIU Yi-cen;YANG Lin;WANG Yang-ning;YIN Cheng-feng;GUO Yu-jun;ZHANG Xue-qin(Electric Power Research Institute of State Grid Sichuan Electric Power Company, Chengdu 600072,China;State Grid Sichuan Electric Power Company, Chengdu 610041,China;School of Electrical Engineering, Southwest Jiaotong University, Chengdu 611756, China)

机构地区：[1]国网四川省电力公司电力科学研究院,四川成都600072 [2]国网四川省电力公司,四川成都610041 [3]西南交通大学电气工程学院,四川成都611756

出　　处：《电工电能新技术》2022年第3期55-62,共8页Advanced Technology of Electrical Engineering and Energy

基　　金：国网四川省电力公司科技项目(52199720003B)。

摘　　要：绝缘子污闪是电力系统长期运行需面临的传统问题,而在积污阶段准确评估绝缘子污秽程度,可合理安排清扫周期,为污闪的防治提供理论指导。本文利用高光谱与红外图像技术的优势,建立了基于信息融合的绝缘子污秽程度检测模型,为绝缘子污秽程度检测新方法提供了新思路。通过人工涂污制备了不同污秽程度的绝缘子进行耐压试验,采集了染污绝缘子高光谱及红外图像。利用图像处理技术,对采集的图像进行校正、去噪、分割等操作,有效降低环境、仪器噪声等因素的干扰。并通过特征提取技术,得到了能够有效表征绝缘子污秽程度的高光谱与红外图像特征量,将不同的特征量进行融合建立了基于高光谱与红外图像信息融合的绝缘子污秽程度检测模型,实现从不同角度对绝缘子污秽程度进行评估,模型的检测准确率为90%。该方法突破了传统检测方法的瓶颈,可为输电线路绝缘子污秽程度智能运检提供新的思路。Pollution flashover of the insulator is a traditional problem to the power system for a long time.Accurate evaluation of insulators pollution degree in the pollution accumulation stage can arrange the cleaning period reasonably,provide theoretical guidance for the prevention and control of pollution flashover.In this paper,an insulator pollution degree detection model based on the advantages of hyperspectral and infrared image technology is established,which provides a new idea for insulator pollution degree detection.Insulators with different pollution degrees were prepared and voltage withstand tests were carried out.Then hyperspectral and infrared images of insulators were collected and the image processing technology was used to correct,denoise and segment the collected images which can reduce the interference of environment and instrument noise effectively.Through the feature extraction technology,the feature quantities of hyperspectral and infrared images which can effectively represent the pollution degree of insulators were obtained.An insulator pollution degree detection model based on the fusion of hyperspectral and infrared image characteristic quantities was established.The results show that the pollution degree of insulators can be evaluated from different angles and the detection accuracy of the model is 90%,which breaks the bottleneck of traditional detection methods.Consequently,this method can provide a new idea for the intelligent inspection of the pollution degree of insulators on transmission lines.

关 键 词：污秽程度 瓷绝缘子 高光谱成像技术 红外成像技术 信息融合 

分 类 号：TM216[一般工业技术—材料科学与工程]