Neuro-fuzzy hybrid control system of nonlinear process in petroleum plant

This paper proposes a practical control method using neural networks and fuzzy control techniques. Neural networks effectively simulate a well-experienced operator's procedure to control the tank level with estimation of a rough control target. Fuzzy control techniques compensate the estimated rough control target using operator's knowledge. The control system was applied to a reflux tank level control of a hydro-desulfurizing plant in the feed oil switching and it controlled the level effectively.