抄録
This paper describes a technique to select promising counter actions by considering operationality and rough effect estimation from the ones derived based on a functional plant model. The plant model is composed of (1) hierarchical relations among goals and functions, (2) plant abstract structure, and (3) component information on behavioral and operational aspects. The knowledge of operationality and effect of a counter action is extracted from plant experts in various anomaly scenarios. Effect estimation is also carried out considering arrangement and rough behavior of components. Reasonable selection results are obtained by the proposed technique in an application to an oil refinery plant.
