MIMO system identification of a slab reheating furnace

This paper describes an application of a multivariable system identification procedure to estimate the parameters of a dynamic model for a walking beam type reheating furnace at the Hoogovens' hot strip mill. The new control philosophy based on actual slab temperatures requires a different controller design by developing a black box model using multivariable system identification (MIMO). Practical knowledge of the furnace process and a priori knowledge of physical properties of heating processes are combined to choose a simple model structure, identical per tracking zone-i.e. a combination of a top and bottom burner zone. During several sessions, data from the process were collected and analysed. Using a multivariable parameter estimation procedure, the ARX method, the parameters per zone were estimated. Due to the mainly horizontal heat flow through the furnace, the structure could be chosen in such a way that by connecting the input/output of the individual zone models, a complete furnace model was obtained. This rather simple dynamic model proved to give a good representation of the temperature characteristics in the furnace.<<ETX>>