Three-Dimensional Geometric Digital Twin of Cable-Stayed Bridges: UAV- and TLS-Based Point Cloud Registration, Fusion Modeling, and Damage Inspection

Advanced inspection techniques are essential for the efficient detection of damage, particularly in long-span bridges situated over challenging terrains, such as rivers. This study introduces an innovative methodology for reconstructing three-dimensional geometric digital twin (DT) models and conducting associated damage inspections on large cable-stayed bridges. The proposed approach utilizes a multi-source registration technique, integrating point cloud data acquired from two types of unmanned aerial vehicles and terrestrial laser scanners. The proposed framework is implemented on an extra-large cable-stayed prestressed concrete bridge located in Jiamusi, China. Three distinct registration methods are employed to achieve fusion modeling of the point cloud data following noise reduction. Experimental results indicate that the automatic registration algorithm significantly improves the absolute accuracy, relative accuracy, and integrity of the DT model by 28.2%, 10.8%, and 18.8%, respectively. Furthermore, the enhanced DT model facilitates the detection of various forms of bridge damage, including surface concrete spalling and deformations in the deck and girders, thereby providing a promising solution for effective bridge maintenance planning.