Estabelecimento de estratégias de controle inteligente na laminação de produtos planos

Abstract

In this doctoral thesis we propose a new neural control strategie for the hot strip rolling mill process. The control objective is to minimize strip thickness variation and the strip crown, wich is the main atribute for a good quality product. The sytem model, extremely complex, is presented in a mathematical formulation and is used as a base for a simulation tool. This tool may be used to analyse, compare and validate general control strategies for the thicknes control problem. As a result we show that adaptive strategy presents better performance then a convencional strategy, with real data, kindly given by Siderar S.A., a Argentine steel making plant.