基 于 点 云 数 据 的 大 型 复 杂 空 间 结 构 数 字 孪 生施 工 监 测 技 术

Due to the multitude of components and intricate joint structures in large complex space structures, traditional modeling methods are inefficient and computationally demanding, failing to meet the needs of such extensive structures. To address this problem, a 3D laser scanning-based digital twin method for monitoring large complex space structures is introduced to explore reverse modeling techniques. Initially, 3D laser scanning captures the structural point cloud. The complete dataset is then aligned using a full datum registration and the ICP algorithm, followed by noise reduction through statistical filtering and voxelization. For beams and columns with different cross-sectional forms, the RANSAC-based centerline fitting algorithm is proposed to extract the component axes) for the component connecting joints, the nearest-neighbor-point algorithm is proposed to carry out the axial end point fitting. Finally, the geometric accuracy and finite element analysis of the digital twin model are verified. The results show that the digital twin model based on point cloud data has high consistency with the BIM design model in terms of accuracy and finite element results. This verifies the accuracy of the digital twin technology in geometric mapping and performance prediction, and lays the foundation for the application of this technology in construction monitoring.