Predictive control of a reheating furnace using a detailed model

Abstract

A model-based control system has been developed for a continuous steel slab reheating furnace with the objective of enhancing heating quality and energy efficiency. The control methodologies employed deliberately use a detailed high-fidelity numerical model. This base model is cited from an earlier study. The present study is dedicated to a control design that encompasses linear, but primarily nonlinear predictive, strategies to optimize heating. The high-order and nonlinear nature of the model has been specifically treated to allow its use in predictive control. To achieve this, a custom model reduction method based on balanced truncation has been applied. This reduction method maintains the structure of the model, including the individual slabs and refractory walls, and preserves the nonlinearities, making the reduction transformation applicable in any state. The reduction transformation is employed to enable tractability for simulating the predictive model, which is piece-wise linearized around the current predicted trajectory. The provided simulation demonstrates the applicability and computational feasibility of predictive control with a complex model.