Grey Bayesian network model for reliability analysis of complex system

Complex systems and their components usually have various performance states and the reliability parameters are normally uncertain. Modeling theories that are developed on the basis of binary outcomes and precise reliability information lack sufficient abilities to describe the above phenomena. In this paper, grey system theory and Bayesian network are employed to analyze the reliability of complex system. First, interval grey number is applied to represent the performance state as well as the conditional probability, which can avoid the loss of important reliability information. Second, the intervals of reliability characteristic parameters such as fault rate and posterior probability are obtained with Bayesian network inference and grey global optimization algorithm. Afterwards, vulnerable components and probabilities of possible states can be identified by using comparison rules of interval grey numbers, which is conducive to reliability analysis and fault diagnosis of complex system. Finally, a case about civil aircraft hydraulic system is studied, showing that the proposed approach is effective and convenient for reliability modelling and analysis of multi-state and uncertain systems.