Bayesian network approach for dynamic fault tree with common cause failures and interval uncertainty parameters

Traditional fault tree analysis often assumes that the basic events are independent and the failure parameters are known. Therefore, it is powerless to deal with the correlation among basic events and the uncertainty of failure parameters due to the small failure data. Therefore, a framework based on continuous-time Bayesian network is proposed to evaluate the reliability of fault tree with common cause failures (CCF) and uncertainty parameters. Firstly, the best-worst method (BWM) and hesitant fuzzy set (HFS) are introduced to address the issue of β-factor being influenced by experts’ subjectivity. Then, the interval theory is introduced to deal with the uncertainty parameters. Based on continuous-time Bayesian network, the conditional probability functions of logic gates (i.e. AND gate, OR gate, spare gate, priority AND gate) with CCF are derived, and the upper and lower bounds of failure probability of top event can be solved. Finally, the fault tree of CPU system is given to verify the effectiveness of the proposed framework.