Abstract
Construction sites are particularly susceptible to the effects of extreme weather, with unsafe items posing a significant risk of causing substantial damage to construction projects and neighboring communities. Furthermore, data regarding the materials, machinery, and buildings present at the site are frequently obtained through manual inspection or on-site photography before the advent of extreme weather conditions. This process is resource-intensive and time-consuming. The core innovation of this study lies in the integration of digital twin technology with RandLA-Net-based point cloud semantic segmentation, optimized specifically for construction site safety management under extreme weather conditions. To achieve systematic disaster preparedness for construction sites, this study explores the potential of utilizing three-dimensional (3D) point cloud technology in construction site management. This involves acquiring location information about materials and machinery on construction sites through the development of a construction site point cloud identification system. This system is designed to identify and analyze potential risk factors in the digital twin model of a construction site, thereby optimizing the site layout at all stages. Furthermore, it enables practitioners to rapidly identify, locate, and assess potential risk factors on-site, facilitating the prompt and effective implementation of measures to prevent extreme weather.