摘要
为研究气体绝缘金属封闭开关设备(gas insulated substation,GIS)腔体内部的金属微粒运动特性,建立真实且比例合理的模拟仿真模型,提出一种基于深度学习中的视觉分析方法,通过高速摄像机获取运动过程的视频,使用YOLOv5目标检测模型进行运动金属微粒的检测和DeepSort模型对其进行跟踪,获取腔体内金属微粒的运动状态等数据信息。在实验数据基础上,进行运动特性的分析,如金属微粒的轨迹分布,速度分布直方图与分布规律,速度大小与所经过位置的相关性等。实验结果表明,金属微粒运动范围较广,速度较小,金属微粒的速度大小与不同的位置存在不同程度的相关性。研究与分析的结果可为实际GIS腔体内部金属微粒的检测与运动特性分析提供一定的理论与方法指导。
In order to study the motion characteristics of the moving metal particles inside the gas insulated substation(GIS)cavity,a real and reasonable simulation model was established.A visual analysis method based on deep learning was proposed.The video of the motion process was obtained by high-speed camera,and the YOLOv5 model and DeepSort model were used to detect and track the moving metal particles.The motion state of metal particles in the cavity was obtained.On the basis of test data,the motion characteristics were analyzed,such as the trajectory distribution of metal particles,the velocity distribution histogram and distribution law,and the correlation between the velocity and the passing position.The experimental results show that the particle motion range is wide and the velocity is small,and the velocity of the particle has different degrees of correlation with different positions.The results of the research and analysis can provide some theoretical and methodological guidance for the detection and motion characteristics analysis of metal particles in GIS.
作者
刘宝林
熊永平
李杰
师伟
LIU Baolin;XIONG Yongping;LI Jie;SHI Wei(School of Computer Science,Beijing University of Posts and Telecommunications,Beijing 100876,China;State Grid Shandong Electric Power Research Institute,Jinan 250003,China)
出处
《山东电力技术》
2023年第12期1-10,共10页
Shandong Electric Power
基金
国家电网有限公司科技项目“融合高速光学成像与局放信号检测的GIS设备内部金属异物检测技术研究”(5500-202116132A-0-0-00)。
关键词
GIS
运动金属微粒
视觉分析
运动特性
GIS
moving metal particles
visual analysis
motion characteristics
作者简介
刘宝林(1999),男,硕士在读,主要研究方向为计算机视觉、新型电力系统;熊永平(1982),男,副教授,主要研究方向为计算机视觉、新型电力系统;李杰(1986),男,高级工程师,长期从事开关类设备状态检测、状态评价工作;师伟(1987),男,高级工程师,主要研究方向为高电压与绝缘技术,从事开关类设备状态评价工作。
