1996 Volume 8 Issue 6 Pages 1073-1086
It is required to deal with the unvertainty information about process abnormal knowledge and running data, the heuristic experience of specialist, and the fuzziness of language for knowledge representation for developing an actual process fault diagnostic system. This paper presents an approach to construction of a process fault diagnostic system by use of fault tree analysis and fuzzy reasoning. The production rules are constructed on the basis of the information about minimal cut sets of a fault tree and the fault propagation in the process. The fuzziness about the consequent part in a rule is determined by proposing fuzzy failure analysis, and stored in the knowledge base as the form of failure possibility distribution. We apply fuzzy reasoning to fault diagnosis by matching the pattern of the grades of process operation data with the knowledge base. The positive causes and their diagnostic grades of process abnormality are estimated from the results of fuzzy reasoning. Finally, we show and verify the practivality of the proposed method by applying the diagnostic system to a laboratory plant.
