DIVA: Recognition of typical situations for turbine generator diagnosis

DIVA is an expert system for turbine generator diagnosis. Its aim is to help a plant operator interpret vibration evolution to diagnose developing faults. Interpretation of vibration evolution is primarily a recognition process which is achieved by a particular model of heuristic classification. DIVA tries to recognize the observed situation among a hierarchy of typical situations.The purpose of this paper is to present the model we used for heuristic classification, based on prototypes. Prototypes represent typical problem situations and may also be seen as experts with special knowledge of what to do in such situations. We argue that this model presents significant advantages to deal with complex problems such as vibration surveillance.