Vision-based geometric finite element model updating for cable suspension bridges

The geometric characteristics of a structure are crucial in determining the accuracy and reliability of finite element (FE) models. However, conventional FE models, which are typically based on theoretical specifications from the design stage, often fail to account for geometric discrepancies introduced during fabrication and construction processes. To address this limitation, this study proposes a methodology that employs computer vision techniques to refine geometric inconsistencies in FE models using point cloud data. The proposed method achieves a closer match between the FE model and the actual target structure by aligning it with point cloud data and refining nodal coordinates. A simulation test was conducted to ensure that the updated FE model accurately represents the geometric characteristics of the target structure, and its practical applicability was validated through an experimental test under actual conditions. This method could further enhance the applicability of FE models in civil engineering, and is expected to contribute to the development of state-of-the-art solutions for computer vision-based geometric FE model updating.