Abstract
The heating section of an annealing furnace is a plant that raises the temperature of steel strips to the desired target temperature to ensure that the strips achieve the desired material properties. Model predictive control (MPC) has been used to increase the temperature in previously published studies, because it reflects the geometrical and material characteristics of the strip. An accurate temperature predictive model for the annealing furnace is required for the optimization of the MPC. For a large and complex annealing furnace, nonlinear models from existing studies are computationally expensive. Therefore, we propose identification method of a linear low-order model, and then apply it to the MPC. In addition, we introduce parameter estimation and an adaptive observer for the estimation to address difficulties in identifying unknown parameters. To verify the identified model and adaptive observer-based MPC, control results from the tracking of target temperature are shown and analyzed through a simulation with conditions close to the actual operation conditions. Furthermore, the simulation results of the proposed method are compared with those of the PI controller and nonlinear MPC. The proposed method solves the problems caused by the massive computational complexity and absence of sensors. The proposed method allows for the accurate control of the temperature using MPC in various types of annealing furnaces.
Highlights
Annealing operation is a process of heat treatment in which the temperature of steel strips increases rapidly and decreases slowly in the cold rolling process [1], [2]
SIMULATION CONDITION As shown in Fig. 7, the highly accurate nonlinear model was assumed as a plant [5], [12] and the model was reduced by coarser spatial discretization for computational reason
This study proposes the identification of a low-order model of annealing furnace and the design of an adaptive observer-based Model predictive control (MPC) using the identified model to control the annealing furnace’s strip temperature with different strip profiles, such as ds, bs, and Tref
Summary
Annealing operation is a process of heat treatment in which the temperature of steel strips increases rapidly and decreases slowly in the cold rolling process [1], [2]. The larger the size of HS, the more sensors are required to measure more parameters; in practice, the number of the sensors cannot be adequate to minimize sensor installation cost or owing to the installation environment These differences in the HS can cause the following problems when obtaining a nonlinear model for the MPC [4], [5]. We present MPC for temperature control of a large HS with the identified low-order model with a small number of states. The computational complexity problem caused by large and complex HS is solved by replacing the nonlinear model to the identified low-order model; the number of states decreases, which enables the MPC controller to achieve low computation time. J x Vector of states y Output (Tsy) ζ Process noise η Measurement noise
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
