Seismic response prediction for nonlinear isolation bridges using output only

Abstract In earthquake-prone regions, the rapid post-earthquake evaluation of numerous bridges poses a significant challenge, further exacerbated by infrastructure aging. While monitoring technologies are actively being developed to address this issue, sensor placement and measurement selection remain critical obstacles, as bridges often feature unique design conditions that require generalized analysis methods and exhibit strong nonlinear behavior during earthquakes. Such behavior restricts the applicability of methods relying on linear indicators and typically necessitates direct measurement of earthquake accelerations. This study proposes a method for predicting the seismic response of nonlinear isolation bridges without measured earthquake accelerations, using only a limited number of sensors installed on the bridge (formulated as an output-only joint state–input estimation problem). The approach integrates nonlinear observability analysis of the structure–sensing system with Bayesian state estimation. A case study on a typical seismic isolation bridge demonstrates the applicability and effectiveness of the method. Results show that, when observability conditions and appropriate estimator settings are satisfied, nonlinear structural responses and input seismic accelerations can be reliably reconstructed. The proposed method enhances the cost efficiency of sensing technologies and provides valuable insights for broader practical implementation.