ARMA Model for Tracking Accelerated Corrosion Damage in a Steel Beam

This paper proposes an enhanced vibration-based damage detection index leveraging autoregressive moving average (ARMA) time-series modeling. The method relies on the fact that material deterioration alters the vibration features of the structure. Thus, the proposed method employs an innovative usage of the ARMA time-series modeling to capture subtle shifts in the vibration response. Specifically, first, a reference ARMA model is fitted on the acceleration response of the undamaged structure. Next, a damage index (DI) is built from the goodness of fit between predicted responses from the reference ARMA model and the actual measured damaged-state acceleration data. Experimental validation was conducted on a steel beam subjected to a controlled accelerated corrosion (up to 40% thickness loss), simulating real-world degradation. Accelerations due to quick-release tests were collected using two accelerometers, along with thickness measurements providing ground-truth damage progression. Results demonstrate that the proposed method can provide sufficient sensitivity in detecting early-stage corrosion progression. This finding highlights the proposed usage of ARMA model's potential for early structural damage detection, offering significant advantages for safety and maintenance strategies in civil engineering applications.