Controle automático de adição de AlF3 em cubas de redução de alumínio, usando lógica fuzzy

Abstract

This work describes a data analysis system, which generates an intelligent aluminum fluoride (AlF3) addition strategy for electrolytic cells. The project is based on Fuzzy logic concepts, wherein the knowledge accumulated by the process specialist is translated in a qualitative manner into a set of linguistic rules such as: IF<condition> THEN<action>. The use of this fuzzy technique for fluoride addition control seeks to explicitly represent an implicit knowledge, retained by the cell operators of the electrolytic cells. Due the conventional system not to contemplate the variations of the phenomena that they involve the dynamics of the process, a controller fuzzy was implemented in the real system for taking of decisions, using the minimum model of Mandani. Based this model, the process variables to input fuzzy system, such as bath temperature and fluoride of aluminum were manipulated to estimate the tendency up and down, through the Method of Least Squares(MLS). The controller fuzzy is applied to calculate the amount of fluoride of aluminum (AlF3) to be added in the electrolytic cells in an automatic way without the need of the the specialist?s of the process intervention. The use of a fuzzy control system is justified by the fact of not having an available dynamic model for aluminum fluoride addition process in electrolytic cells. This lack of modeling is due to the great complexity of the phenomena involved in a cell, which are thermodynamic and electromagnetic processes linked together.