Prediction of parameters of a fluid catalytic cracking plant - FCC using neuronal models

Abstract

This document presents the design ofneural models that can predict some parameters of afluid catalityc cracking unir, located in an European refine/y, using historical data. In thiskind ofwork, it is necessary to take information about the behavior ofthe system, in this casethe behavior ofthe FCC unit, which is usedfor neural network training. This work propasesa methodolagy for system identification and parameter prediction in processes with a highlevel of complexity, such that linear modeling techniques are not enough.