改变施工安全管理模式的虚实整合系统:设计、开发与部署

Recently, sensing technologies have been widely proposed and applied in construction jobsite safety monitoring, providing volumetric and various data. However, existing research focuses on utilizing parts of the data for a specific application, failing to provide a comprehensive picture of a jobsite safety situation based on the big data. To solve this problem, this research proposes a context aware construction jobsite management system based on cyber physical system. First of all, the project aims to propose a distributed data computing and communication architecture for safety management big data analysis; then the project aims to identify the data requirements; then the project investigate the big data analysis technologies for safety management data analysis; next, data visualization techniques is used to construct a context-aware virtual environment based on sensing data and processed information; finally, a differential information feedback model is proposed. The project has five main contributions to the body of knowledge: 1) it propose a distributed architecture for safety management big data computing and communications; 2) it identifies the required data for construction jobsite safety management, providing to a comprehensive data information network; 3) it attempts to apply deep learning for safety management; 4) it proposes a framework of organizing and visualizing the needed data and information in the virtual environment to create a context-aware scenario for facilitating the jobsite safety decision makings; and 5) It identifies the differential data/information requirements of workforces on site and proposes an information feedback system deployment model based on internet of things and augmented reality. The safety management system prototype can also utilize the big data to provide a big picture on jobsite safety for the safety management.professionals, improving the safety management productivity and therefore enhancing the jobsite safety performance.

近年来，传感器在施工现场的广泛应用为施工安全提供了大数据，在一定程度上提高了现场管理水平。但目前的研究多集中利用局部数据作特定应用(如高空坠落预警)，未能充分挖掘数据价值、将施工现场与虚拟环境进行有机结合并提供直观全面的决策辅助。为解决此问题，项目将设计及开发连接施工现实环境与建筑虚拟模型的虚实整合系统。首先根据网络硬件及约束条件，提出四层分布式大数据计算与传输架构；对施工安全管理实践进行分析，确定系统需求；在系统的不同层面，设计包括深度学习在内的算法对现场采集的数据进行处理，将信息在虚拟现实中可视化，为安全管理人员提供决策支持。将差异性数据反馈给现场施工人员。本项目理论贡献在于全面分析了施工现场安全管理数据需求，提出符合移动计算约束的网络架构，探讨了深度学习在施工安全管理中的应用，建立了现场与虚拟世界有机结合的双向数据传输与可视化模型。

近年来，传感器在施工现场的广泛应用提供了施工安全大数据，一定程度上提高了现场管理水平。目前研究多集中利用局部数据作特定应用(如高空坠落预警)，未能充分挖掘数据价值、将施工现场与数字环境进行有机结合并提供直观全面的决策辅助。本项目对施工现场安全管理的数据采集、处理及表达方式进行了研究。首先根据网络硬件及约束条件，提出四层分布式的基于传感器的现场安全管理系统架构；设计了融合结构化及非结构化数据的工人安全风险评估模型；对施工安全管理实践进行分析，确定工地安全管理系统的数据需求及加工处理过程；使用虚拟现实对工人的安全行为特点进行了研究。最后对于安全信息的表达方式对安全人员决策认知影响进行了实验研究，设计及测试了安全信息在虚拟环境中的反馈方法。项目研究成果为基于数字孪生的施工安全研究与应用奠定了理论基础。项目中采用的研究方法及范式，也能应用到人机交互界面设计及研究，并扩展应用在其他行业。