Uncertainty Propagation on Seismic Responses of a Bridge Pier Using Bayesian Active Learning

Key Points

• This research aims to evaluate the impact of uncertainties on the seismic responses of bridge piers using an efficient Bayesian active learning method.
• Developed a Bayesian active learning technique based on Gaussian process regression sampling.
• Numerically derived statistical estimates and uncertainties of seismic response values.
• Formulated learning functions and stopping criteria for the analysis.
• Applied the method to analyze uncertainties in maximum and residual displacements of RC bridge piers.
• Demonstrated effective efficiency in estimating posteriors of seismic response probability distributions compared to traditional methods.
• Validated the proposed method's effectiveness through comparisons with Monte Carlo simulation results.