配电线路表面外破缺陷小目标双通道YOLO识别算法  

Double Channel YOLO Recognition Algorithm for Small Targets With Surface External Damage Defects in Distribution Lines

作　　者：李海波 付家兴 颛俊尧 刘永亮 LI Hai-bo;FU Jia-xing;XU Jun-yao;LIU Yong-liang(State Grid Inner Mongolia Eastern Electric Power Co.,LTD.Chifeng Power Supply Company,Chifeng 024000)

机构地区：[1]国网内蒙古东部电力有限公司赤峰供电公司,赤峰市024000

出　　处：《环境技术》2024年第7期111-117,共7页Environmental Technology

摘　　要：配电线路的外破缺陷可能导致电气事故、电弧故障和设备损坏,但外破缺陷识别易受波动因子影响,导致识别效果不佳。为此,本文提出一种基于配电线路表面外破缺陷小目标双通道YOLO识别算法。根据传输线理论对线路网络进行分布参数表示,并重建三维点云;结合双通道YOLO算法,压缩特征信息,计算波动因子;通过簇聚类分析获取线路参数分布的特征点,并在临时坐标系中求取缺陷判别函数;将符合判别函数的特征匹配点像素聚合区域作为缺陷区域,实现缺陷检测。实验结果显示,所提方法应用后,识别查全率显著较高,线路表面外破缺陷识别精度更高。The external damage defects of distribution lines may lead to electrical accidents,arc faults,and equipment damage,but the identification of external damage defects is easily affected by fluctuation factors,resulting in poor identification results.For this purpose,this article proposes a dual channel YOLO recognition algorithm based on small targets of surface external damage defects in distribution lines.Based on transmission line theory,represent the distribution parameters of the line network and reconstruct a three-dimensional point cloud;Combining the dual channel YOLO algorithm to compress feature information and calculate fluctuation factors;Obtain characteristic points of line parameter distribution through cluster clustering analysis,and obtain defect discrimination function in temporary coordinate system;Using the pixel aggregation area of feature matching points that meet the discrimination function as the defect area to achieve defect detection.The experimental results show that after the application of the proposed method,the recognition recall rate is significantly higher,and the accuracy of identifying external damage defects on the line surface is higher.

关 键 词：配电线路 外破缺陷 小目标双通道YOLO 识别精度 

分 类 号：TM246[一般工业技术—材料科学与工程]