A METHOD OF QUALITATIVE REASONING FOR MODEL-BASED PROBLEM-SOLVING ANDITS APPLICATION TO A NUCLEAR-PLANT

Model-based expert systems are expected to contribute to overcoming th e difficulties of conventional rule-based systems. This paper describe s the modeling of mechanical systems and a method of qualitative reaso ning based on causal specifications. The causal specifications represe nt a component's local causal properties, following the principles for reusability and composability. it contributes to providing intuitive causal ordering of complex behavior originated in the combination of c omponents, including inter-component negative feedback. A model of a s ystem is represented by combining a set of local component models and global knowledge derived from general properties of the physical entit y. This allows for reusable knowledge which is easy to describe. Furth ermore, the method has been successfully applied to a nuclear power pl ant. Reasoning results were unambiguous and matched those obtained by domain experts. Copyright (C) 1996 Elsevier Science Ltd