A coupled knowledge based system using fuzzy optimization for advisory control

Abstract

Abstract A knowledge based system is described that is designed to generate on‐line advice for operators regarding the proper distribution of hydrogen resources in a refinery. The system uses a coupled architecture incorporating numerical computing in a knowledge based system environment. This arrangement allows for powerful and flexible problem solving. One portion of the coupled system formulates an optimization problem that is subsequently solved by an external routine. This application is particularly concerned with uncertainty that is present in some of the constraints. To deal with this uncertainty, a fuzzy approach to the optimization is taken. A method is presented that solves the fuzzy optimization problem using standard mathematical programming techniques. The results of the fuzzy optimization allow the crisp solution to be expanded into a neighborhood of solutions that is considered acceptable. Although this work examines a specific problem, the concepts presented are general.