An improved method for system fault diagnosis using fault tree analysis

This paper proposes an improved method to diagnose system faults using fault tree analysis to optimize system diagnosis. All minimal cut sets, their occurrence probabilities and the diagnosis importance of components were determined via fault tree analysis used for system reliability. The order by which the minimal cut sets were checked depended on the diagnosis importance. Components with larger diagnosis importance within a cut set were checked first. Meanwhile, the minimal cut sets containing the component with the largest diagnosis importance were also diagnosed first when the component was fault. According to the diagnostic sequence of system components, a diagnostic decision tree was generated. This method offsets the insufficiency caused by the diagnostic sequence which is merely determined by the diagnosis importance of components. Therefore, the minimal cut set which has larger occurrence probability can be checked first, and the system diagnosis is optimized. An example is given to demonstrate the efficiency of this method.