Bayesian Networks for Calculating the Diagnostic Importance Factor (DIF) in the Presence of Correlated Failure

Fault diagnosis and System reliability in the presence of correlated component failure (CCF) is the main challenge in new complex system design. In this paper we use Bayesian networks for reliability analysis and then the diagnostic importance factor is presented. The correlation between failures of the components ismodeled by encoding multivariate Bernoulli distributions process, and then the process of diagnosing is considered by developing a Bayesian network. The approach is illustrated through common redundant structures, k-out-of-n and Parallel-series systems. The results indicate that the approach is simple and efficient. Finally we show comparison between our proposed method and the previous methods in which the CCF is not considered.

Keywords:

systems reliability , correlated component failures , Bayesian networks , System diagnosis