Lukas Bitschnau, Martin Kozek, Stefan Jakubek, and Florian Leeber
Process Control, Predictive Control, Kalman filtering, Continuous Annealing
This paper presents a successful implementation of a model predictive controller (MPC) in an continuous annealing furnace, based on a compact process model with few physical parameters. The complete furnace structure, including the conveyed strip, the heater elements, and the thermal effects of the cladding, is therefore described by a discrete state space model of low order. To ensure stability of the discretized process model a physically motivated sufficient stability criterion is presented. For state estimation, an efficient Kalman filter algorithm is introduced. A model based predictive controller, which also respect considers process constraints, is customized to the special application of the continuous annealing furnace. Experimental results demonstrate the effective industrial implementation and the potential of constrained model based predictive control.
Important Links:
Go Back
