Reliability analysis of complex system based on dynamic fault tree and dynamic Bayesian network

Abstract

Traditional static fault tree analysis is widely used to analyze the reliability of complex systems in different fields. To improve their reliability and availability values of complex redundant systems, lots of dynamic gates are used, such as Priority AND (PAND) Gate, Spare Gate, Sequence Enforcing (SEQ) Gate and Functional Dependency (FDEP) Gate. And dynamic fault tree developed on the basis of Markov chain is applied. In order to reduce calculation and avoid finding minimal cut set, dynamic Bayesian network is introduced. And then methods to convert dynamic fault tree events into corresponding Bayesian network nodes are put forward and conditional probability tables are determined by domain experts and logic relations between nodes. At last, an aviation electric system is taken for example. According to its dynamic fault tree model, dynamic Bayesian network model is established, and expanded from the first time slice to the second time slice. The results show that the reliability of aviation electric system decreases gradually when there is no repair. And it will maintain at a high level when repair measures are taken. Through importance analysis, weak nodes in design are pointed out.