基于边缘计算的核电工程有限空间作业安全管理研究

Research on Edge Computing⁃based Safety Management of Confined Space in Nuclear Power Engineering

针对核电工程有限空间安全管理所面对的感知数据量大、数据传输信号差、数据安全性弱等问题,设计了边云结合的核电工程有限空间作业安全管理框架,改变传统依赖集中云平台进行数据分析的计算范式。在该框架基础上,结合实际项目调研,设计了核电工程有限空间作业安全预警系统。该系统以智能布控球配合嵌入式系统作为边缘端,采用轻量化MobileNet作为目标识别模型。测试结果表明,该系统能有效处理有限空间视频数据,目标识别精度超过98%,识别模型的运行速率为62帧(FPS)、运行时延为16ms,能满足核电工程有限空间作业安全管理的实际需求。

In order to solve the problem of large amount of sensory data,poor data transmission signal and weak data security of safety management in the confined space of nuclear power engineering,a safety management framework for the confined space in nuclear power engineering is designed to change the traditional computing paradigm that relies on centralized cloud platform for data analysis.On the basis of this framework,a safety warning system for the confined space in nuclear power engineering is designed for actual projects.The system uses an intelligent deployment sphere with an embedded system as the edge end and a lightweight MobileNet as the target recognition model.The test results show that the system can effectively process the video data of the confined space,the recognition accuracy exceeds 98%,and the recognition model runs at 62 frames per second(FPS)with a running time delay of 16ms,which can meet the safety management requirements of confined space in nuclear power engineering.