Credible Bayesian reliability model for structures with interval uncertain parameters

Abstract

If a non-probabilistic interval model is adopted to describe the uncertainty, the determination of the interval bounds is challenging, which degrades the accuracy and credibility of the reliability analysis. In this study, based on Bayesian theory, a non-probabilistic credible Bayesian reliability model, which can be updated by introducing new sample points, is proposed for three failure modes and developed to model the credibility of non-probabilistic reliability. The relations between the classical probabilistic confidence reliability model and the proposed non-probabilistic credible Bayesian reliability model are investigated and discussed. Two relevant numerical examples are selected for the investigation to demonstrate the effectiveness of the proposed model and the limitations of the classical confidence reliability in the evaluation of structural safety.