Design of a Self-Tuning Expert Controller for a Gasoline Refinery Catalytic Reformer

The objective is to explore design concepts for self-tuning expert controllers. To accomplish this, two interacting rule based controllers for supervisory control and system optimization are constructed to control a gasoline catalytic reformer. The knowledge bases for the controllers are established from human operator experience and basic engineering knowledge about the process dynamics. Inference is provided by a fuzzy logic engine. After manual tuning of the expert controller scaling coefficients is accomplished, a crisp heuristic is developed for self-tuning. The performance of the self-tuning expert controller is tested against perturbations of a simulation model of the catalytic reformer.