Knowledge-Based Decision and Conversation in a Hybrid Diagnosis System

Abstract In the framework of a hybrid fault diagnosis system it is possible to embody expert knowledge about the target technical system structure and behaviour. Such a diagnosis system follows the computational model of the human diagnostician, based on soft computing and observations from sensors and human operators, too. It combines neural and symbolic knowledge processing, automatic and interactive diagnosis. This paper presents the decisional and conversational block, called Intelligent Diagnosis Finish (IDF). It is the feed-back loop that includes the process operator in the iterative diagnosis procedure, in order to utter new observable manifestations, when a set of inconsistent neural-network hypotheses results.