Performance Assessment of Friction Pendulum Bearings in Bridges via Data Assimilation

Seismic isolation techniques are widely used in regions susceptible to earthquakes. From a long-term maintenance perspective, it is crucial to develop methodologies for assessing the post-installation performance of these systems based on monitored data. Data assimilation techniques provide a versatile framework for addressing various challenges in complex environments by estimating the posterior distribution of unknown system states, enhanced by prior physical knowledge of systems. This study explores the feasibility of assessing the performance of friction pendulum bearing systems (FPSs) in bridges using data assimilation. Specifically, dynamic data from a scaled bridge model in large shaking table tests and from an actual bridge under earthquake conditions are utilized. Both an iterative Bayesian approach and a batch Bayesian approach are applied to address time-variant and time-invariant parameter estimation challenges. The results demonstrate that the state of the bridges and their FPSs can be accurately evaluated under diverse input conditions.