基于链式贝叶斯网络的结构可靠性分析

Structural reliability analysis based on chain-like Bayesian network

将贝叶斯网络与传统可靠性方法结合,建立结构系统的可靠性贝叶斯网络模型,通过改进的分支限界法确定结构主要失效模式,并将贝叶斯网络链式化来提升计算效率。根据可靠性方法计算条件概率表;使用概率网络估算法来考虑主要失效模式之间的相关性,计算系统可靠性;当有新信息出现时,利用贝叶斯网络推理,对结构系统可靠性进行评估。以一桁架结构为研究对象,计算结构系统的可靠性,并在新信息出现的情况下对系统可靠性进行了更新。

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.