A Novel Fault Diagnosis Framework Based on Reliability Analysis and Arithmetic Circuit

To meet the real-time diagnosis requirements of complex systems, this paper presents a novel fault diagnosis framework based on dynamic fault tree and arithmetic circuit. It pays attention to meeting two challenges: model development and real-time reasoning. Specifically, we use a dynamic fault tree to model dynamic fault modes and calculate some quantitative parameters using algebraic technique and Bayesian network (BN) in order to avoid the state space explosion problem. Furthermore, we compile a BN into an arithmetic circuit to obtain answers to probabilistic queries by evaluating and differentiating the arithmetic circuit. In addition, we incorporate sensors information into diagnosis process and propose the schemes on how to update the diagnostic importance factor and the minimal cut sequences. Finally, the example of a train-ground communication system is used to demonstrate the proposed method.