Quality Defect Detection of Distribution Network Engineering Based on Lightweight Improved YOLOv5

被引：0
作者：
Yang L. [1 ]
Wang J. [1 ]
Duan X. [1 ]
Li J. [1 ]
Li Y. [1 ]
Li F. [1 ]
机构：
[1] China Electric Power Research Institute, Haidian District, Beijing
来源：
Dianwang Jishu/Power System Technology | 2023年 / 47卷 / 09期
关键词：
defect detection; depthwise separable convolution; distribution network; engineering quality; lightweight improvement;
D O I：
10.13335/j.1000-3673.pst.2022.2063
中图分类号：
学科分类号：
摘要：
The distribution network faces the end power users and is a critical infrastructure serving the national economy and the people's livelihood. The quality of the distribution network project directly affects the safety and reliability of the power supply. The traditional manual inspection has a low efficiency and high risks, which cannot meet the requirements of the digital and intelligent management and control of the distribution network. The common mobile terminals, such as the unmanned aerial vehicles, the robots and the smartphones, are limited in their computing powers, and the chips are restricted in their storage spaces in the embedded devices. Therefore, it is a difficult technical problem to be solved to improve the defect identification accuracy of the distribution network project with the limited computing resources. Based on the depthwise separable convolution, this paper improves the BottleNeck design, a high multiplexing module in the YOLOv5 network, compresses the network parameters and computation of the YOLOv5 with different depths and widths by about 30% - 50%, and realizes the lightweight transformation of the network. The experimental scheme is designed, and the actual distribution network engineering quality data set is used for testing and comparison. The typical application cases are analyzed, which verifies the effectiveness of the improved method proposed in this paper. The improvement of defect detection speed and accuracy are realized and a strong support is given for the improvement of the digitalization and intelligence level of distribution network engineering management and control. © 2023 Power System Technology Press. All rights reserved.
引用
收藏
页码：3864 / 3872
页数：8
相关论文
共 32 条
[1]  
LI Xuefeng, LIU Haiying, LIU Gaohua, Transmission line pin defect detection based on deep learning[J], Power System Technology, 45, 8, pp. 2988-2995, (2021)
[2]  
GU Chaoyue, LI Zhe, SHI Jintao, Detection for pin defects of overhead lines by UAV patrol image based on improved faster-RCNN [J], High Voltage Engineering, 46, 9, pp. 3089-3096, (2020)
[3]  
GAO Wei, Chen ZHOU, GUO Moufa, Insulator defect identification via improved YOLOv4 and SR-GAN algorithm[J], Electric Machines and Control, 25, 11, pp. 93-104, (2021)
[4]  
JIA Xiaofen, YU Yeqi, GUO Yongcun, Lightweight detection method of self-explosion defect of aerial photo insulator[J/OL], High Voltage Engineering, pp. 1-7, (2022)
[5]  
ZHAO Zhenbing, LI Yanxu, QI Yincheng, Insulator defect detection method based on dynamic focus loss function and sample balance method[J], Electric Power Automation Equipment, 40, 10, pp. 205-211, (2020)
[6]  
Bin LI, LI Yalin, ZHU Xinshan, Multi-target detection in substation scence based on attention mechanism and feature balance [J], Power System Technology, 46, 6, pp. 2122-2131, (2022)
[7]  
XUE Minghua, AI Chunmei, Intelligent image processing technology for safety helmet wearing in power plant[J], Proceedings of the CSEE, 42, 9, pp. 3346-3353, (2022)
[8]  
LIU Haiying, MO Wenhao, TAN Yuanpeng, Oriented object detector for transmission line equipment detection[J], Power System Technology, 45, 12, pp. 4888-4895, (2021)
[9]  
ZHANG Jun, WANG Jiye, SONG Rui, Research on efficient detection technology of transmission line abnormal target based on edge intelligence[J], Power System Technology, 46, 5, pp. 1652-1661, (2022)
[10]  
ZHANG Ye, GAO Yuhan, HUANG Xinbo, Fault identification and evaluation of tower wading in 110~220kV transmission line with fine mask RCNN[J], Power System Technology, 46, 6, pp. 2132-2144, (2022)
1234