基于视频数据的火电厂作业人员安全行为实时分析的研究

Research on Real-time Analysis of Safety Behavior of Thermal Power Plant Operators Based on Video Data

火电厂属于高风险环境,实时分析作业人员安全行为对保障人员安全和火电厂的运行至关重要。为此,对作业人员是否吸烟,是否佩戴安全帽,是否在危险区域滞留等安全行为的预警设计了一款实时的火电厂作业人员安全行为分析系统。针对火电厂作业人员安全行为检测方法易受环境的干扰,存在漏检与误检问题,本文改进了YOLOv5算法,具体来说,在YOLOv5进行特征融合时,采用了注意力门控单元,注意力门控单元能够显著提高安全行为检测能力,同时此模块参数较少,满足实时性需求。最后,在SHWD和吸烟行为数据集上验证了所提的方法。根据实验表明,本方法能够显著提高安全行为检测能力,能够应用于实时的火电厂作业人员安全行为分析系统中。

Thermal power plant is a high-risk environment.Real-time analysis of operator safety behavior is very important to ensure personnel safety and the operation of thermal power plant.To this end,a real-time safety behavior analysis system for thermal power plant operators is designed for early warning of safety behavior such as whether the operators smoke,wear safety helmets,and whether they stay in dangerous areas.Aiming at the problem that the safety behavior detection method of power plant operators is susceptible to environmental interference,and there are problems of missed detection and false detection,this paper improves the YOLOv5 algorithm.Specifically,when YOLOv5 is used for feature fusion,the attention gating unit is used,which can significantly improve the safety behavior detection ability,and this module has fewer parameters to meet real-time requirements.Finally,the proposed method was verified on the SHWD and smoking behavior datasets.The experimental results show that this method can significantly improve the ability of safety behavior detection,and can be applied to the real-time safety behavior analysis system of power plant operators.