摘要
变电运维工作对变电站安全稳定运行起着至关重要的作用。随着变电站数量的不断增长,设备运维工作面临着工作量激增与人员相对短缺的突出矛盾,导致不能及时发现设备缺陷以及人员违规作业。针对此现状,文章提出了一种基于YOLOv4的多路视频实时检测技术,通过多个摄像头对变电站进行全方位多角度的监控,基于目标检测网络YOLOv4对变电站内的人员作业行为以及设备缺陷等28种类型进行实时检测,通过量化压缩模型以及优化基于实时流传输协议(real time streaming protocol,RTSP)的推流方法显著提升了多路视频检测的实时性和稳定性。目前,基于YOLOv4的多路视频实时检测装置已经在大连某变电站成功部署运行,极大地保障了变电站的运维安全。
Substation operation and maintenance work plays a vital role in the safe and stable operation of substations.With the continuous increase in the number of substations,the equipment operation and maintenance work is facing the prominent contradiction between the surge in workload and the relative shortage of personnel,resulting in failure to discover equipment defects and personnel irregular operations in time.In view of this situation,this paper proposes a multi-channel video real-time detection technology based on YOLOv4,which can monitor the substation from all directions and angles through multiple cameras.Based on the object detection network YOLOv4,28 types of personnel operation behaviors and equipment defects in the substation are detected in real time.By quantifying the compression model and optimizing the RTSP(real time streaming protocol),the real-time performance and stability of multichannel video detection is significantly improved.At present,the multi-channel video real-time detection device based on YOLOv4 has been successfully deployed and operated in a substation in Dalian city,which greatly ensures the operation and maintenance safety of the substation.
作者
靳敏
廖逍
王兴涛
邱镇
JIN Min;LIAO Xiao;WANG Xingtao;QIU Zhen(State Grid Information&Telecommunication Group Co.,Ltd.,Beijing 102211,China)
出处
《电力信息与通信技术》
2022年第12期32-38,共7页
Electric Power Information and Communication Technology
基金
国网信息通信产业集团有限公司科技项目资助“变电站多路视频与声纹智能识别技术研究及系统开发”(5268002XXX38)。
