Decision Trees Generation Based on Fault Trees Analysis

摘要

This paper proposes a fault tree analysis method to generate diagnostic decision trees. All minimal cut sets, their occurrence probabilities and components' diagnosis importance factors are determined via fault tree analysis used for system reliability. Minimal cut sets represent minimal sets of component failures that cause a system failure, so the diagnostic sequence of minimal cut sets should be first assigned. Minimal cut sets with larger occurrence probabilities are diagnosed first. The order by which components within each minimal cut set are depend on the diagnostic importance factor ordering; components of larger diagnosis importance factor are checked first. Moreover, if the component which has the largest diagnosis importance factor within the minimal cut set being checked is failed, the minimal cut sets including it are also diagnosed first. According to diagnostic sequence of system components, a diagnostic decision tree can be generated. This method avoids deficiency caused by the diagnostic sequence which is merely determined by the diagnosis importance factors of components, regardless of occurrence probabilities of minimal cut sets. This paper presents an example to demonstrate efficiency of this method.