Algoritmos genéticos para sintonia simultânea de múltiplos controladores em processos de refino

Abstract

This work proposes the use of genetic algorithms to tuning decoupled controllers for multivariable systems for refine process. It is presented a new fitness function for genetic algorithm that considers both ITSE (Integral Time Squared Error) and minimum variance criteria. The proposed technique can be applied to tune different control architectures also including non-linear controllers to process presenting strong interactions among its variables. In order to demonstrate the performance of the proposed method, it is applied to three multivariable processes Wood-Berry Distillation Column, Isopropanol Distillation Column and Fluid Catalytic Cracking (FCC), with the use of PID (proportional-integralderivative) and PD-fuzzy decentralised controllers. This approach shows good performance and can be extend to different kinds of controllers and processes.