Automatic Diagnosis via Sensors Modeled by Dynamic Fault Trees

This paper presents a new automated diagnosis methodology which attempts to bridge the gap between reliability at the design phase and diagnosis at the usage phase. The methodology takes advantage of dynamic fault tree qualitative and quantitative data to develop a diagnostic importance measure. The methodology produces a diagnostic decision tree based on the fault tree and on the diagnostic importance measure. To enhance the diagnosis process the presented methodology incorporates evidence from sensors to improve diagnosis.