Efficient Bridge Geometric Digital Twinning Using a Parametric Bridge Generator

Abstract

Geometric digital twinning is an essential step for the holistic storage, analysis, visualization, and interpretation of information about bridges from multiple modalities. Generally, accurate digital twinning depends on high-quality sensing data and the information extracted from such data, whereas the acquisition of high-quality data is typically guided by the information embedded in the model. This data–model coupling suggests that data-based modeling tasks (DBMT) and model-based data tasks (MBDT) should be performed iteratively using the available information at each stage. This research investigates an efficient bridge geometric digital twinning using a parametric bridge generator (PBG) that can accommodate DBMT and MBDT at different levels encountered during field visual inspection. Flexibility and efficiency can be achieved by (1) the PBG generating a model in the photorealistic synthetic environment once key parameters are obtained in the field, (2) flexibly selecting the key parameters depending on the level of accuracy and detail, and (3) enabling the PBG to generate as reasonable a model as possible, even with missing parameters to guide the inspection process. Based on the PBG, this research develops two feasible DBMT and MBDT that can guide bridge inspection and the geometric digital twinning process, model-based data evaluation (MBDE), and data-based model evaluation (DBME). These evaluations are employed to clarify the degree to which modeling requirements are satisfied for a given level of detail. The proposed PBG-based geometric digital twinning approach is demonstrated through two case studies in the field, one for short-span road bridge and one for long-span cable-stayed bridge.