Control of a distillation column using fuzzy inference systems

Distillation is an essential separation process in the chemical and petrochemical industry. Keeping a constant product composition in a distillation column is of great importance from an economical point of view. In this paper, we use fuzzy inference systems (FISs) to control the products´ compositions in a binary distillation column. In order to adjust the FIS conclusion parts, a reinforcement learning method, a fuzzy Q-learning (FQL) method, is used. Results obtained show that FISs are able to satisfactorily perform the control task, while complying with the given specifications