基于深度学习的架空输电线路绝缘子缺陷检测方法研究综述

Overview on Defect Detection Methods of Insulator for Overhead Transmission Lines Based on Deep Learning

绝缘子是架空输电线路中不可或缺的部件,对其进行定期检修能确保电力的安全传输和电网的安全运行。人工巡检、机器人巡检、载人直升机巡检、无人机巡检等是现有的输电线路巡检方式。目前,我国电力线路运维的主流模式是“无人机巡检为主,人工巡检为辅”。为了安全起见,操作无人机飞行检修高压输电线路时,必须与线路保持一定的安全距离。由于绝缘子在无人机拍摄的输电线路图像背景复杂多变且状态复杂,小目标种类占比较多。故本文针对架空输电线路绝缘子缺陷检测的场景,分析了目标检测算法的常见类型,并比较了不同算法策略的优缺点,结合实际应用对算法进行改进,最后展望绝缘子缺陷检测的研究趋势。

Insulator is indispensable component in high voltage transmission lines.Regular maintenance of insulator can assure safe transmission of power and safe operation of power grid.Manual patrol inspection,robot patrol inspection,manned helicopter patrol inspection and unmanned aerial vehicle patrol inspection are the existing patrol inspection methods of transmission lines.At present,the mainstream mode of operation and maintenance of power line in China is“UAV patrol inspection as the main mode,supplemented by manual patrol inspection”.For the sake of safety,the UAV,in case of inspecting the HV transmission line,must keep a certain safe distance from the transmission lines.Since the image background and the state of the insulator in the transmission line taken by the VAV is various and complex and also there are more types of small targets,the common types of target detection algorithm for the scenario of detection of insulator of transmission line is analyzed in this paper,the pros and cons of different algorithm strategy is compared and the algorithm is improved in combination with actual application.Finally,the research trend for the defect detection of insulator is prospected.