摘要
读懂服务器面板上图标及指示灯信号对运维人员来说是一项特殊的工作技能,面对复杂多样性的服务器图标和指示灯组合信号,提出了一种基于视频图像识别的服务器故障辅助诊断系统。首先建立服务器面板图形图标指示灯的故障特征库、进行模型训练和模型验证;采集故障视频、运用OpenCV这一计算机视觉库对视频进行逐帧解析,提取连续的关键帧,分析灯光颜色,并与故障特征库比对,综合故障特征给出诊断结论,并将诊断结果反馈给终端用户。
It is a special work skill for person to understand the icon and indicator signal on the server panel.Since there are redundant sever icons and complex flash assembly,a diagnostic system based on video image recognition was developed.The operation methods of the system are listed as follows:Firstly,establish the fault feature database of the server panel graphic icon indicator,perform model training and model verification;collect fault video,use OpenCV computer vision library to parse the video frame by frame,extract continuous key frames,analyze the color of the light,and analyze The fault signature database is compared,the comprehensive fault feature is given a diagnosis conclusion,and the diagnosis result is fed back to the end user.
作者
唐琳
王顺阳
王永刚
张天冉
董月娇
Tang Lin;Wang Shunyang;Wang Yonggang;Zhang Tianran;Dong Yuejiao(Dezhou Power Supply Company,State Grid Shandong Electric Power Company,Shandong Dezhou 253000,China;Yucheng Power Supply Company,State Grid Shandong Electric Power Company,Shandong Yucheng 251200,China;YGsoft Inc.,Guangdong Zhuhai 519085,China)
出处
《科技通报》
2020年第8期58-61,86,共5页
Bulletin of Science and Technology
基金
国网山东省电力公司科技项目(编号:5206081800Y7)。
