铁道科学与工程学报2025，Vol.22Issue(4)：1870-1882,13.DOI:10.19713/j.cnki.43-1423/u.T20240833
一种用于接触网腕臂绝缘子破损大小检测的YOLOv7-OpenCV算法
A YOLOv7-OpenCV algorithm for detecting the size of insulator damage in overhead contact system
摘要
Abstract
To address the problem that traditional machine-assisted manual fault detection methods for catenary wrist arm insulators cannot achieve intelligent damage level identification and real-time monitoring,a YOLOv7-OpenCV algorithm was proposed to realize real-time quantification of damage size detection.The proposed algorithm employed an adaptive histogram equalization(AHE)strategy to improve the local contrast of the detected image,introduced a global attention module(GAM)and an omni-dimensional dynamic convolution(ODConv)module into the YOLOv7 backbone network to enhance the extraction of key feature edge information such as damage shape and size,and improved the model's attention and detection accuracy for insulator damage targets.Then the OpenCV algorithm was integrated into the improved YOLOv7 algorithm.Through filtering,edge detection,and contour extraction algorithms,the damaged contour was accurately identified,achieving precise quantification of the size of insulator damage.Through comparative analysis of the detection performance between on-site and experimental images,the results show that the proposed algorithm has an average detection accuracy of 95.43%for damaged targets,which is 6.67%higher than the original YOLOv7 algorithm.This indicates that the accuracy of damage recognition is improved.The detection rate is 41.3 frames per second,satisfying the minimum frame rate requirement for real-time monitoring.Compared to the OpenCV algorithm,the proposed algorithm reduces the average error rate of the damage size recognition by about 10%.This indicates that the proposed algorithm exhibits satisfactory detection accuracy and efficiency.The proposed real-time quantitative algorithm for precise positioning and damage size detection of wrist arm insulators is beneficial for improving the intelligent inspection level of traction power supply.关键词
腕臂绝缘子/检测/YOLOv7/OpenCV/破损大小/量化Key words
wrist arm insulators/detect/YOLOv7/OpenCV/damage size/quantification分类
交通工程引用本文复制引用
张廷荣,段承成,胡延文,张建鑫,范振华..一种用于接触网腕臂绝缘子破损大小检测的YOLOv7-OpenCV算法[J].铁道科学与工程学报,2025,22(4):1870-1882,13.基金项目
兰州交通大学青年科学基金资助项目(1200061169,1520020821) （1200061169,1520020821）
