Integrating Photogrammetry and Topological Data Analysis within a Digital Twin Framework for Missing Bolt Detection in Bridges

Abstract

Abstract. Reliable inspection of bridge infrastructure is essential for maintaining structural safety, particularly in identifying missing bolts that may compromise system performance. This study represents an extension of our previous work with a methodology that integrates point cloud-based Digital Twin (DT) models with Topological Data Analysis (TDA), to enable accurate detection and localization of missing bolts. A high-resolution 3D representation of bridge joints is first generated using a photogrammetric reconstruction process. YOLOv8 is then employed to detect and localize bolt positions within the point cloud data. Subsequently, Alpha complexes are utilized within the TDA framework to capture topological features and identify anomalies associated with missing bolts. The approach is validated through a benchmark case study, demonstrating high accuracy in detecting missing bolts and robustness to variations in point cloud density. The results indicate that the integration of DT and TDA provides an effective and reliable solution for advanced structural health monitoring applications.