Bayesian approaches for evaluating wind‐resistant performance of long‐span bridges using structural health monitoring data

Reliable estimation of wind-induced displacement responses long-span bridges is critical to evaluating their wind-resistant performance. In this study, two Bayesian approaches, generalized linear model (BGLM) and sparse learning (SBL), are proposed for characterizing the lateral with structural health monitoring (SHM) data. They fully model-free data-driven preferable reckoning total intended performance assessment. With measured wind speeds, a BGLM developed characterize nonlinear relationship between response speed, where class selection (BMCS) criterion incorporated determine optimal model. formulation by SBL, both speed direction treated as explanatory variables elicit probabilistic structure. The SBL cleverly makes resulting exempt from overfitting generalizes well on unseen formulated models then utilized forecast in extreme typhoon events beyond scope, predicted contrasted finite element analysis results design maximum allowable under serviceability limit state (SLS). methods demonstrated using data acquired GPS sensors anemometers instrumented suspension bridge. show that superior prediction amenable SHM-based evaluation conditions.