Structural reliability analysis based on chain-like Bayesian network

This paper discusses a new computational framework for structural reliability assessment while traditional reliability method can't handle structure system with uncertain information. The Bayesian network approach is combined with the improved branch-and-bound methods to simplify the Bayesian network model. Turn the converging BN to the chain-like BN to improve its efficiency and to facilitate its application to the large structures. The system failure probability is computed with the PNET while the correlations between the failure sequences are considered. The structure reliability assess with the BN inference when new information was available. A truss structure with 10 bars is analyzed to illustrate the proposed method.