中国电力 ›› 2019, Vol. 52 ›› Issue (7): 31-39.DOI: 10.11930/j.issn.1004-9649.201806102

• 泛在电力物联网——先进信息与通信技术 • 上一篇    下一篇

基于YOLOv2网络的绝缘子自动识别与缺陷诊断模型

赖秋频, 杨军, 谭本东, 王亮, 傅思遥, 韩立伟   

  1. 武汉大学 电气工程学院, 湖北 武汉 430072
  • 收稿日期:2018-06-26 修回日期:2018-12-07 出版日期:2019-07-05 发布日期:2019-07-13
  • 作者简介:赖秋频(1995-),男,硕士研究生,从事电力系统图像处理与模式识别研究,E-mail:846472189@qq.com
  • 基金资助:
    国家自然科学基金资助项目(51277135)。

An Automatic Recognition and Defect Diagnosis Model of Transmission Line Insulator Based on YOLOv2 Network

LAI Qiupin, YANG Jun, TAN Bendong, WANG Liang, FU Siyao, HAN Liwei   

  1. School of Electrical Engineering, Wuhan University, Wuhan 430072, China
  • Received:2018-06-26 Revised:2018-12-07 Online:2019-07-05 Published:2019-07-13
  • Supported by:
    This work is supported by National Natural Science Foundation of China (No.51277135).

摘要: 针对无人机或机器人获取的输电线路绝缘子图像,提出了一种基于深度学习图像识别框架(YOLOv2)网络的输电线路绝缘子在线识别与缺陷诊断模型,训练YOLOv2网络学习复杂背景下各种绝缘子的特征并准确识别,结合边缘检测、直线检测、图像旋转和垂直投影方法,对识别出各种状态的绝缘子进行缺陷诊断。输电线路巡检图像的仿真结果表明,所提出的绝缘子自动识别与缺陷诊断方法能迅速准确地从输电线路巡检图像中识别出绝缘子,并诊断出绝缘子是否破损以及缺陷位置,有利于提升输电线路智能巡检水平。

关键词: 输电线路, 智能巡检, 绝缘子, YOLOv2网络, 深度学习, 图像识别, 缺陷诊断

Abstract: Aiming at the transmission line insulator images obtained by drones or robots, this paper proposes an online recognition and defect diagnosis model of transmission line insulators based on YOLOv2 network. The YOLOv2 network is trained to learn and accurately recognize the characteristics of various insulators under complicated background, and eventually to accomplish the defect diagnosis of the identified insulators of various status by means of edge detection, line detection, image rotation and vertical projection methods. The simulation results of the patrol inspection images of the transmission lines show that the proposed automatic insulator identification and defect diagnosis method can quickly and accurately identify the insulators from the patrol images of the transmission lines and diagnose the defects and their locations of the insulators, which is beneficial to enhance the intelligence inspection level of transmission lines.

Key words: transmission line, intelligent inspection, insulator, YOLOv2 network, deep learning, image recognition, defect diagnosis

中图分类号: