Intelligent Fault Diagnosis Method Based on Reliability Analysis

This paper presents a novel fault diagnosis method which overcomes the shortcomings of traditional approaches. It uses dynamic fault tree model for reliability analysis. All minimal cut sequences are generated via a new modular method, while the diagnostic importance factor (DIF) of components and minimal cut sequences are calculated using discrete-time Bayesian network. Also, we introduce the cost of diagnostic importance factor (CDIF) for components to evaluate the influences of the test costs, and combine it with minimal cut sequences' DIF to determine the order of the system diagnosis. Finally, an example is given to demonstrate the effectiveness of this method.