基于AI视频技术的水电厂设备不安全状态自动化预警研究  

Research on Automatic Warning of Unsafe State of Hydroelectric Power Plant Equipment Based on AI Video Technology

作　　者：袁璞 王冠琪 宋刚伟 何超 YUAN Pu;WANG Guanqi;SONG Gangwei;HE Chao(State Grid Shaanxi Ankang Hydropower Plant,Ankang 725000,China)

机构地区：[1]国网陕西安康水力发电厂,安康725000

出　　处：《自动化与仪表》2024年第6期132-136,共5页Automation & Instrumentation

摘　　要：为及时发现设备的不安全状态,预防潜在的安全隐患,降低设备故障的风险,提出了基于AI视频技术的水电厂设备不安全状态自动化预警方法,确保水电厂稳定运行。利用摄像机采集水电厂设备的AI视频图像,并增强AI视频图像,提升图像清晰度;通过在轻量型YOLOv5算法提取增强AI视频图像的特征;通过在预测框筛选机制内,引入得分惩罚机制,结合提取的特征,预测水电厂设备的不安全状态;通过声音预警形式,对不安全状态预测结果进行自动化预警。实验证明,该方法可有效实时采集水电厂设备的AI视频图像,并增强AI视频图像;可精准预测水电厂设备的不安全状态,可有效自动化预警设备不安全状态,并呈现预警级别与时间等信息。To timely detect the unsafe status of equipment,prevent potential safety hazards,and reduce the risk of equipment failure,an AI video technology based automated warning method for unsafe status of hydropower plant equipment is proposed to ensure stable operation of the hydropower plant.Using cameras to capture AI video images of hydroelectric power plant equipment,and enhancing AI video images to enhance image clarity.Extract enhanced features from AI video images using the lightweight YOLOv5 algorithm.By introducing a score penalty mechanism within the prediction box filtering mechanism and combining the extracted features,the unsafe state of hydropower plant equipment is predicted.Automated warning of unsafe state prediction results through sound warning.Experimental results have shown that this method can effectively collect real-time AI video images of hydropower plant equipment and enhance AI video images.It can accurately predict the unsafe status of hydropower plant equipment,effectively automate the warning of equipment unsafe status,and present information such as warning level and time.

关 键 词：AI视频技术 水电厂设备 不安全状态 自动化预警 摄像机 轻量型YOLOv5 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]