无人机人工智能图像识别技术在国网基建作业中的应用.pdfVIP

长，信息通信

DOI:10.20153/j.issn.2096-9759.2025.03.035

无人机人工智能图像识别技术在国网基建作业中的应用

陈广朋，李帅1，吴冈

刚,王森2

1国网河北省电力有限公司建设公司，河北石家庄050000；

2河北电力工程监理有限公司，河北石家庄050000

ApplicationofDroneArtificialIntelligenceImageRecognitionTechnology

inStateGridConstructionOperations

CHENGuangpeng,LIShuai,WUGang?,WANGSen?

1StateGridHebeiElectricPowerCo.,Ltd.ConstructionCompany,Shijiazhuang050000,China;

2HebeiElectricPowerEngineeringSupervisionCo.,Ltd.,Shijiazhuang050000,China

摘要：为国网基建作业中的科学管理和Abstract:Toprovidescientificmanagementanddecision-makingsupportforStateGridcon-

决策提供依据，确保国网基建项目按期、structionprojccts,ensuringthatStateGridconstructionprojectsarecompletedonscheduleand

保质保量顺利跋工。首先以YOLOv3-SEofhighquality.Firstly,adroneimageautomaticrecognitionsystembasedontheYOLOv3-SE

算法为主要技术构建无人机图像自动识algorithmisconstructedasthemaintechnology,andthentheSE-blockattentionmoduleisinte-

别系统，然后在YOLOv3算法的基础上gratedintotheYOLOv3algorithmtorealizethedesignoftheYOLOv3-SEalgorithm.Thepre-

融入SE-block注意力模块实现YOLOv3-cisionofthedroneimageautomaticrecognitionsystemisevaluatedbythemAPvalueevaluation

SE算法设计。通过mAP值评价法对无method;finally,theequipmentstatusofpowerpoles,conductorconnections,clamps,linechan-

人机图像自动识别系统的精准度进行评nels,andsubstationequipmentisvisualizedthroughdronepatrol,andthelinkagevisualization

价；最后通过无人机巡航对电力杆塔、导oftheoverallequipmentstatusoftheprojectisdetermined.Theaverageprecisionofpicture

线连接点、耐张线夹、线路通道以及变电recognitionmAPvalueisverifiedtobe94.5%.Inthisway,artificialintelligencehascompleted

站设备等完成国网基建作业设备状态可thepreciserecognitionoftargetphysicalobjectsinStateGridconstructionoperations,andthe

视化，并确定环节预警分析逻辑和项目systemperformanceis