基于监控视频图像的智慧变电站远程可视化自动巡检方法  

Remote Visualization and Automatic Inspection Method of Smart Substation Based on Surveillance Video Image

作　　者：徐佳 胡宗义 XU Jia;HU Zongyi(Nanrui Group(State Grid Electric Power Research Institute)Co.,Ltd.,Nanjing,Jiangsu 210000,China)

机构地区：[1]南瑞集团(国网电力科学研究院)有限公司,江苏南京210000

出　　处：《自动化应用》2024年第15期240-242,共3页Automation Application

摘　　要：由于变电站数据采集受各种磁场干扰,采用传统巡检方法得到的可视化巡检结果较模糊。为此,提出了一种基于监控视频图像的智慧变电站远程可视化自动巡检方法。首先建立变电站监控视频图像数据集,采用图像数据增强技术扩充数据集,然后通过巡检定位技术识别变电站中的关键设备和区域,最后,基于YOLOv3网络框架实现巡检图像识别以及可视化自动巡检功能。结果表明,应用所提方法,其巡检结果准确率均高于95.3%,优于对比方法,具有良好的巡检效果。Due to the interference of various magnetic fields in the data collection of substations,the visual inspection results obtained by traditional inspection methods are relatively vague.To solve this problem,this paper proposes a remote visual automatic inspection method for smart substations based on monitoring video images.Firstly,establish a video image dataset for substation monitoring,expand the dataset using image data augmentation technology,identify key equipment and areas in the substation through inspection positioning technology,and finally,implement inspection image recognition based on the YOLOv3 network framework to achieve visual automatic inspection function.The experimental results show that the accuracy of the inspection results using the proposed method is higher than 95.3%,which is better than the comparison method and has a good inspection effect.

关 键 词：监控视频图像 智慧变电站 远程可视化 可视化自动巡检 

分 类 号：TM631[电气工程—电力系统及自动化]