Reliability analysis of dynamic reliability blocks through conversion into dynamic bayesian networks

Recently, the dynamic Reliability Block Diagram (DRBD) with the characteristics of intuitiveness and visibility has been extensively employed in industrial fields, such as aerospace, military facilities, nuclear energy and chemical industry, etc. However, there exists a limitation in the past research that the complexity of the reliability calculation considering the additional modelling constructs, such as state dependency (SDEP) block, spare part (SPARE) block, load sharing (LSH) block and so on. To address the issue, this paper proposes the dynamic Bayesian networks (DBN) based on converting the DRBD into a DBN. Based on these rules, the constructs of DRBD can be mapped into DBN and easily analyzed leveraging the mature algorithms and tools. A case study of a multiprocessor distributed computing system is adopted to illustrate the conversion algorithm. The results show that this approach can signally increase the analytic efficiency of DRBD for critical large-scale system.