基于边缘技术和PowerNet网络的输电线缺陷检测  

Defect Detection on Power Lines Based on Edge Technology and Deep Network

作　　者：陆晓 吴强 蒋承伶 马洲俊 王茂飞 单华[3] LU Xiao;WU Qiang;JIANG Cheng-ling;MA Zhou-jun;WANG Mao-fei;SHAN Hua(State Grid Jiangsu Electric Power Co.,Ltd.,Nanjing 210024,China;State Grid Taizhou Power Supply Company,Taizhou 225300,China;Jiangsu Fangtian Electric Power Technology Co.,Ltd.,Nanjing 211100,China)

机构地区：[1]国网江苏省电力有限公司,江苏南京210024 [2]国网江苏省电力有限公司泰州供电分公司,江苏泰州225300 [3]江苏方天电力技术有限公司,江苏南京211100

出　　处：《光学与光电技术》2024年第5期45-55,共11页Optics & Optoelectronic Technology

基　　金：国网江苏电力有限公司科技项目资助(J2022004)。

摘　　要：针对传统输电线检测易受大雾天气影响、检测精度和效率低等问题,提出了一种融合多模块去雾网络、网络分区策略和PowerNet网络模型的输电线路高精度、高效率检测方法.首先,在端对端学习的基础上,设计了一种多模块融合的去雾网络,解决了因雾导致输电线缺陷检测精度低的问题.然后,为了提高边缘技术的巡检效率,设计了一种基于二进制粒子群的网络分区策略.进而提出PowerNet网络模型来解决输电线缺陷检测精度低的问题.最后,通过实验对所提出的方法进行效果验证和数据分析,实验结果表明:该方法具有较高的缺陷检测准确率和实时性,其精度和效率分别可以达到 99.3%和 28 ms/张,具有较高的工程实用价值.The traditional transmission line detection is easy to be affected by foggy weather,and the detection accuracy and efficiency are low.In this paper,a high-precision and high-efficiency detection method is proposed,which integrates multi-module de-fog network,network partitioning strategy and PowerNet network model.Firstly,based on the end-to-end learning,a multi-module defogging network is designed to solve the problem of low detection accuracy of transmission line defects caused by fog.Then,in order to improve the inspection efficiency of edge technology,a network partitioning strategy based on binary particle swarm is designed.On the basis of these,PowerNet network model is proposed to solve the problem of low accuracy of transmission line defect detection.Finally,the method proposed in this paper is validated and analyzed by experiments.The experimental results show that the proposed method has high accuracy and real-time defect detection,and its accuracy and efficiency can reach 99.3%and 28 ms photo,respectively.It can be seen that the method proposed in this paper has high engineering practical value.

关 键 词：输电线 缺陷检测 图像去雾 网络分区 检测精度 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]