Abstract
This study presents a methodology for structural health monitoring (SHM) of viaducts using Linear Variable Differential Transformer (LVDT) sensors installed on pot bearings. The aim is to use continuous LVDT measurements to evaluate and predict bearing condition and potential degradation. Two complementary approaches are proposed: (1) condition assessment based on regulatory standards and (2) data-driven anomaly detection techniques. The first approach focuses on accumulated sliding distance; a key parameter derived from LVDT data and is evaluated according to legislation. The second approach analyzes vibration responses during train passages to estimate dynamic properties of the bearings, mainly natural frequencies and damping ratios. This methodology combines signal filtering, Blind Source Separation (BSS), and system identification methods such as the Natural Excitation Technique (NExT) and the Ibrahim Time Domain (ITD) approach. From the identified parameters, time series are constructed and evaluated in real time using control charts and multivariate statistical tools, such as Hotelling’s T 2 . By combining both techniques, this approach enhances the early detection of bearing anomalies and contributes to predictive maintenance strategies. The integration of LVDT monitoring with advanced data analysis provides a robust framework for real-time SHM and improved management of viaduct infrastructure.