基于YOL011与双光图像分析的光伏板积尘检测方法

打开文本图片集

中图分类号：TM615 文献标志码：A 文章编号：2095-2945（2025）25-0001-0

Abstract：Asoneof themostcommerciallyvaluablegreen energysources，solar energyholdsasignificantcompetive advantageinthefieldofpowergeneration.Photovoltaic（PV）modules，beingthecorecomponentsforphotoelectricconversionare pronetodustacumulationontheirsurfacesduringoperation，which blockssolarradiationandleadstoadeclinein photoelectricconversioneficiencyandoutputpower.Existing dustdetectionmethodsbasedonDep LearningAlgorithmsheavily relyontherichnessandqualityofannotateddata.Toadressthisisue，thispaper proposesadustdetectionalgorithmbasedon YOLO11combinedwithhistogramanalysisanddual-lightimagefusion.Firstly，theYOLO11-OBBrotatingobjectdetection algorithmisutilizedtoaccuratelyextractPVstringsandPVmodulesfrombothvisiblelightandthermalimagingimages. Secondly，forindividualPVstrings，theYOLOl1-SEGMENTalgorithmisemployedtodetectdustareasinvisiblelightimages andtoidentifythermalspotsinthermalimagingimages，analyzingthetemperatureofthesespots.Fialytraditionalimage procesingmethodsareappliedtoanalyzethevisiblelightimages，predictingthedustcoverageratiobycombininghistogram features.ThismethodleveragesthepowerfulfeatureextractionandgeneralizationcapabilitiesoftheYOLOlalgorithm， incorporatestraditionalimageproessngmethodsforsecodaryverification，andclasifiestheseverityofdustaccmulationbased onabnormaltemperaturedataofcorespondingcomponentsinthealimagingimages，significantlyimprovingtheacuracyof dust detection in complex scenarios.

Keywords：YOLO11; histogram；photovoltaic dustdetection；dust accumulation severity；photovoltaic module clean

随着太阳能发电技术的快速发展，光伏组件作为太阳能发电系统的核心部件，其运行效率直接关系到整个系统的发电性能。（剩余9972字）