An expert system architectural framework for engineering selection

Abstract

A generic architectural framework for constructing flexible expert systems for engineering selection is proposed in this paper. A two-tiered framework is suggested to mirror the two steps of elimination and refinement in selection tasks. The top layer consists of reverse rules for implementing the elimination of inapplicable alternatives, and the bottom layer consists of multiple-criteria decision-making (MCDM) models for explicitly modeling the decision-making processes in selection. Some MCDM models which involve user preferences or weights are incorporated in the framework because the selection task involves user preferences or weights assigned to decision parameters. The framework has been applied to the domain of solvent selection for carbon dioxide removal processes, and a prototype advisory system based on the framework has been developed. The proposed framework contributes to the field of engineering selection because it enables the explicit representation of selection knowledge, and the formal modeling of user preferences.