Virtual trial assembly (VTA) using 3D laser scanning as the digital carrier can overcome the shortcomings of time‐consuming and costly physical preassembly. However, its application in large steel structures with bolted connections remains limited. First, this study introduces a novel approach for acquiring multiscale point cloud data of large steel members using terrestrial laser scanners (TLSs) and hand‐held scanner (HHS). This approach considers both the global data and the local details of the steel members. Additionally, a precise registration method based on magnetic 3D targets is proposed for multiscale point clouds, which enables the registration accuracy of multisource point clouds to reach submillimeter precision. Subsequently, a novel algorithm for feature point screening is introduced, which utilizes a dichotomous point cloud grid approach to identify and extract a significant quantity of bolt holes effectively. This approach enables fully automated and fast extraction of the point cloud on the cylindrical inner surface of the holes. Finally, the bounding box and Procrustes analysis approach are employed to perform VTA using the point cloud of the cylindrical bolt holes as the assembled features. The accuracy and feasibility of the above method are verified by a down‐scale modeling experiment and project test, which provide technical support for the VTA of large steel truss structures.

中文翻译：

---

基于多尺度点云融合的大型螺栓连接钢构件虚拟试装

采用3D激光扫描作为数字载体的虚拟试装（VTA）可以克服物理预装配耗时且成本高昂的缺点。然而，其在具有螺栓连接的大型钢结构中的应用仍然有限。首先，本研究介绍了一种使用地面激光扫描仪（TLS）和手持式扫描仪（HHS）获取大型钢构件多尺度点云数据的新方法。这种方法考虑了钢构件的全局数据和局部细节。此外，提出了一种基于磁性3D目标的多尺度点云精确配准方法，使得多源点云的配准精度达到亚毫米级精度。随后，引入了一种新的特征点筛选算法，该算法利用二分点云网格方法有效地识别和提取大量螺栓孔。这种方法可以全自动、快速地提取孔的圆柱形内表面上的点云。最后，采用包围盒和Procrustes分析方法，使用圆柱形螺栓孔的点云作为组装特征来执行VTA。通过缩尺建模实验和工程试验验证了上述方法的准确性和可行性，为大型钢桁架结构的VTA提供了技术支撑。