基于优化YOLO算法的电力现场智能安全监控系统研究

打开文本图片集

中图分类号：TP277.2 文献标志码：A 文章编号：1671-5276（2025）05-0299-04

Research on Intelligent Safety Monitoring System for Power Field Based on Optimized YOLO Algorithm

LIU Jianjun （Datang Hebei New Energy（Zhangbei） Co.，Ltd.，Zhangjiakou O7645O，China）

Abstract：To improve thesafetyand managementeficiencyof power workplaces，thisresearch proposesaninteligent safety monitoring system for power sites based on optimized YOLO algorithm.A hybrid feature extraction method was designed，combining background diference methodand interframediferencemethodfortargetpersonneldetection. Feature pyramid network was introduced to optimize YOLOv3 algorithm，and support vector machine was used to classifysecurity behaviors.Theexperimental results showed thatthe average detection accuracy of the detection model was 99.42% ，very effective in detecting behaviors of failing to wear safety helmets，safety belts during high一altitude operations，insulated shoes at thework site，and hang doublehooksduring high-altitude operations，andwith the average recognition acuracy as O.87，0.81，0.84，0.86 respectively，and the average accuracy as0.85.The proposed intelligent safety monitoring system for power fieldcan effectively identifypersonneland monitor their safety behaviors. Keywords： YOLOv3； feature pyramid;electric power facilities；security monitoring； feature fusion

0 引言

电力施工现场安全隐患多发，传统人工巡检方式难以实现全时段、全覆盖的高效监管。（剩余4670字）