生产人员不规范操作多目标检测研究与应用

AResearch and Application of Multi-target Detection for Nonstandard Operation of Operator

随着自动化水平的提高、运行方式的改变,越来越多的厂站采用“无人值班、少人值守”模式。在目前的运行工作中,安全监督监控系统需要运行人员持续高强度地监控各参数,同时借助专业知识对参数进行分析和判断,才能更好地执行监盘的任务,这一定程度上造成了人力和资源的浪费。通过构建智能识别安全监督管理系统,由高清摄像头进行视频图像采集,然后检测服务对视频内容的不规范操作进行多目标检测、分类,可以高效、可靠地解决因值班人员困乏、画面过多等情况,造成值班人员错过某些重要的报警信息,从而对生产或安全造成严重影响,导致生产环境发生很大的安全风险和管理风险。

With the improvement of automation level and the change of operation mode,more and more plants and stations adopt the mode of"unattended and few people on duty"In the current operation work,the safety supervision and monitoring system requires operators to continuously monitor various parameters with high intensity,and at the same time,analyze and judge the parameters with professional knowledge,so as to well perform the task of monitoring,which,to some extent,causes a waste of human resources and resources By building an intelligent identification safety supervision and management system,high-definition cameras will collect video images,and then the detection service will conduct multi-target detection and classification for non-standard operations of video content.Effectively and reliably solve the problem that the personnel on duty are sleepy and have too many pictures,which causes the personnel on duty to miss some important alarm information,thus causing serious impact on production or safety,leading to great safety risks and management risks in the production environment.