改进YOLOv8的高压输电线路轻量化绝缘子缺陷检测算法

打开文本图片集

中图分类号：TM755 文献标志码：A 文章编号：2095-2945（2026）03-0040-04

Abstract：Defectdiagnosisofinsulatorsinhigh-voltagetransmisionlinesisanimportanttaskinpowerinspectionandisof greatsignificancetoensuringthereliabiltyandsafetyofpowertransmision.Thispaperusesconvolutionsharingtoimprovethe detectionheadnetworkofYOLOv8network，andproposesalightweightinsulatordefectdetectionalgorithm.Theshared convolutionaldetectorhadnetworkproposedinthispapercanallowinformationexchangebetweenthethreedetectors，makeful useoftheextractedfeatureinformation，andgreatlyreducetheparameteramount，calculationamount，andweightofthenetwork whileimprovingtheacuracyofmodeldetection.Experimentalresultsshowthattheimprovedalgorithmreducestheparameter amount，calculation amount，and weight by 20% ，19.8 % ，and 21% respectively while ensuring detection accuracy，achieving an effective balance between detection accuracy and lightweight.

Keywords：insulatordefectdetection;lightweight;convolutionsharing;high-voltage transmission lines;detectionaccuracy

随着现代电力系统的规模不断扩大以及智能电网的发展，高压输电线路的安全问题越来越受到重视。（剩余5371字）