ANN-based intelligent control system for simultaneous feed disturbances rejection and product specification changes in extractive distillation process

Abstract

Distillation is one of the most studied processes in the control literature because of its importance as a separation process. However, little attention has been paid to the dynamics and control of the simultaneous changes in the feed and product specifications. In this study, it is proposed an intelligent control system based on artificial neural network for the extractive process to obtain anhydrous ethanol using ethylene glycol as a solvent. The study considered the changes in the azeotropic feed and the specification of anhydrous ethanol simultaneously, taking both the extractive and recovery columns into account, and keeping the process operating at minimum energy consumption condition. The performance of the intelligent control system was evaluated using Aspen Plus Dynamics™, and the results showed that it is is able to efficiently determine the new setpoints of controllers when facing changes in anhydrous ethanol specification and/or disturbances in the azeotrope feed. The new steady-state is reached in a short time interval (2–4 h, depending on the disturbance type). Based on the integral squared error, integral absolute error, and steady-state error, the results showed that the intelligent control system presented superior performance when compared to conventional control systems. The implementation of the developed control is simple and the existing control structure remains unchanged.