Comparison of Methods for Structural Reliability Assessment under Parameter Uncertainties

摘要

Statistical parameters of basic random variables in reliability assessment often can not be determined precisely due to limited sample of data. Uncertainties in parameter estimates are modeled using the Bayesian approach and the confidence interval approach. An interval Monte Carlo method is introduced to propagate interval information of unknown parameters through reliability assessment. Comparisons of the Bayesian approach and the interval approach are illustrated using two examples of finite element-based reliability analyses.