| 摘要 |
<P>PROBLEM TO BE SOLVED: To provide a fault diagnosis system which can determine a fault cause synthetically by reflecting diagnosis results from a plurality of diagnosis models without uniquely determining a diagnosis model even in the case observation data required for determination of a fault type cannot be sufficiently acquired. <P>SOLUTION: A first causal network of the fault diagnosis system stochastically shows a plurality of fault types and relations between respective fault types and observation information having causal relations with the fault types, and a second causal network stochastically shows relations between observation information of fault types and fault causes per fault type, and an observation information acquisition means acquires a plurality of pieces of observation information from a diagnosis object system, and a determination probability calculation means calculates determination probabilities of fault type candidates from the acquired observation information and the first causal network, an occurrence probability calculation means calculates occurrence probabilities of fault cause candidates per fault type from the acquired observation information and the second causal network, and a fault cause informing means informs of a fault cause in accordance with the determination probabilities and the occurrence probabilities of fault cause candidates. <P>COPYRIGHT: (C)2008,JPO&INPIT |
