Controle multivariável aplicado a uma coluna de destilação propano-propeno utilizando a técnica de separações de sinais.

Abstract

The industrial processes are in its majority are multivariable and show a high number of interactions between its variables. Some techniques have been developed to target the reduction of these interactions. Independent Component Analysis (ICA) has been a very promising technique when it is desired to reduce or even eliminate the coupling between variables. The ICA technique has been utilized on the separation of unknown sources. Although there are several different studies targeting the application of ICA in industrial processes, just a few of them geared for technical applications in distillation columns. This dissertation attempts to establish a control strategy applied to a high purity distillation column with vapor recompression using the ICA. In addition, the dissertation proposes to establish a control strategy for the bottom and top of compositions. For this will be a communication between the Aspen Plus DynamicsTM and Simulink / Matlab®. The communication will be established through the AMSimulation block. Two control strategies were compared: MPC with ICA and MPC without MPC. The results of the MPC with ICA strategy were shown to be more promising, showing that the ICA technique is a useful decoupling tool for systems with strong coupling.