基于改进YOLOv7-tiny的变电站场智能管控技术提升

打开文本图片集

中图分类号：MT769 文献标志码：A文章编号：1001-5922（2026）1-0209-04

Abstract：Aiming at the limitations of traditional manual-basedidentityrecognition methods curentlyadopted by substationstaff，this paper proposes anonline intelligent recognition algorithmforsubstation personnel identity based on UAV aerial images.Considering the computing power constraints of on-site UAV terminals and the requirements forreal-time performance and accuracy of inspection results，the algorithm puts forwardacloud-edgecollaborative computing framework.Itdivides the identityrecognition task intotwo stages，namely facedetectionand facerecognition，whicharedeployedontheUAVterminalandthegroundterminal（server）respectively，andrealizesonlinerecognitionof personnel identity through colaborative work.To further reducethe computational complexityat the UAV terminal，theoriginal YOLOv7-tiny model is modified to better adapt to theface detection task insubstations.

Key words ：online identification;inteligent identity recognition;cloud-edge collaboration;detection technology

随着电网智能化水平的不断提升，部分电力场景（输电线路、变电站等）中投入大量的监控设备（固定监控相机、无人机及机器人等）用于日常设备巡视、维护和检修[3]。（剩余6268字）