Modeling and observer design for aluminum manufacturing

摘要

This paper addresses modeling, identification, and observer design of the Hall-Heroult process. This process is the main core of aluminum manufacturing. More specifically, by building upon a blend of physical-chemical constitutive relations and system identification tools, this paper first provides a model describing the interplay of Anode-Cathode Distance (ACD), alumina concentration, and pot pseudo-resistance. The proposed model is then used to tune a linear Kalman filter generating online estimates of the plant state variables. The proposed approach is validated via data taken from a real industrial plant. (c) 2022 European Control Association. Published by Elsevier Ltd. All rights reserved.