Digital Twins (DTs) provide a promising solution for the maintenance and operation of bridges, thanks to their ability to mirror physical/structural conditions. A bridge DT generally consists of a geometric-semantic model whose creation, however, requires extensive manual effort. This paper presents an automated framework to generate the parametric model of bridges from their segmented point clouds. Following the concept of reverse engineering with parametric modeling, Parametric Prototype Models (PPMs) are proposed as tools to extract parameter values from point clouds. A local and global optimization problem is defined to adjust and assemble PPMs into an integrated model. The proposed approach has been validated by applying it to the point cloud of bridge components as well as point clouds captured from six concrete bridges in Bavaria, Germany. The results show that the proposed approach can generate the parametric model of bridges with a mean absolute error (MAE) of 8.71 cm.
Read full abstract