Damage identification of railway bridges through temporal autoregressive modelling

Resumen

The damage identification of railway bridges poses a formidable challenge given the large variability in the environmental/operational conditions that such structures are subjected to along their lifespan. Within a research project devoted to the continuous monitoring of a real in-operation bridge, the Mascarat Viaduct in Alicante (Spain), this paper investigates the potential use of strain sensors and autoregressive (AR) modelling to extract highly sensitive damage features. The viaduct, built in the beginning of the 20th century, belongs to the TRAM line 9 in the province of Alicante (Spain). The viaduct is instrumented with a dense network of strain sensors and accelerometers for monitoring the structural response under train passages. As an initial exploratory investigation, and with the aim of developing a physics-based digital twin of the viaduct, this work presents an introductory numerical assessment of the potential use of strain measurements and AR models for fast damage identification. To this aim, a finite element model (FEM) of the viaduct has been developed to simulate the dynamic response of the viaduct under moving train loads. On this basis, a two-class damage classifier (damaged or non-damaged) is developed by exploiting time series of continuously identified AR parameters. The presented results and discussion evidence the usefulness of the proposed methodology, offering an unsupervised damage detection approach with minimum computational efforts.