摘要
GIS设备出现缺陷会导致变电站发生事故,给电网带来巨大经济损失。针对目前变电站大规模的人工巡检存在着效率低下、易漏检等问题。文中对X射线无损检测得到的GIS设备图片进行研究。对获得的GIS图片进行恰当预处理,提出一种改进的FCM算法进行分割,在传统的FCM分割算法基础上,引入非隶属度、不确定度且结合图像的领域信息进行分割GIS设备图像,采用SVM分类器进行缺陷检测。实验表明,此方法缺陷检测率能达到90%以上。
The defects of GIS equipment will lead to accidents in substations and bring huge economic losses to the power grid.In view of the current large⁃scale manual inspection in substations,there are some problems such as low efficiency,easy to miss inspection and so on.In this paper,the image of GIS equipment obtained by X⁃ray nondestructive testing is studied.Firstly,the obtained GIS images are preprocessed properly,and an improved FCM algorithm is proposed for segmentation.Based on the traditional FCM segmentation algorithm,non membership degree,uncertainty and domain information of the image are introduced to segment GIS equipment images,and finally SVM classifier is used for defect detection.Experiments show that the defect detection rate of this method can reach more than 90%.
作者
李波
胡秀敏
何志琴
高正浩
樊磊
孙博
LI Bo;HU Xiumin;HE Zhiqin;GAO Zhenghao;FAN Lei;SUN Bo(Institute of Electric Power Science,Guizhou Power Grid Co.,Ltd.,Guiyang 550002,China;College of Electrical Engineering,Guizhou University,Guiyang 550025,China)
出处
《电子设计工程》
2021年第1期78-82,共5页
Electronic Design Engineering
基金
中国南方电网有限责任公司科技项目(0066600KK52170036)。
