The fault diagnosis system requires high quality in real-time

accuracy and completeness.Thus

a hierarchical fuzzy signed directed graph(SDG)model was established.Combining with the technique of layering in the fault diagnosis

this model can reduce the searching space and improve the diagnosis efficiency.The fuzzy membership was introduced to carry out quantitative description on fault possibility and meet the needs of fault accurate positioning.According to the qualitative relationship between nodes at failure propagation paths

the hierarchical fuzzy SDG model compatible pathway can be determined.When a fault occurs

the compatible branch will be searched and the candidate nodes of fault source will be found out.The compatibility and fuzzy-reasoning was applied to finally get the possibility of fault sources in order.At last

the Tennessee Eastman process(TEP)was taken as the object to establish the hierarchical fuzzy SDG model and do fault diagnosis reasoning

thus to verify the effectiveness of this method.The results show that this diagnosis method is feasible and the efficiency and accuracy of fault source searching are improved.