Predictive and diagnostic analysis of an holdup tank by means of Dynamic Bayesian Network

In dynamic reliability evaluation, the complete behaviour of the system has to be taken into account. In this paper, a benchmark taken from the literature is examined. To this aim, we exploit Dynamic Bayesian Networks (DBN) extending standard Bayesian networks by introducing a discrete temporal dimension. The goals are the prediction of the system unreliability and the computation of diagnostic indices. Because of the achievement of such goals, we propose DBN to be a valid approach for dynamic reliability evaluation.