Robust predictive controller of the mold level in a steel continuous casting process

The mold level control is one of most important control loops in a steel continuous casting process, since level variations higher than 10^-2m may impair the quality of the steel. The goal of this paper is to apply the Hammerstein Generalized Predictive Controller (HGPC) technique to control the mold level, considering the effects of external disturbs and random noises. The parameters of the controller are tuned by Genetic Algorithms because there is no analytical method for tuning the HGPC to guarantee its stability and robustness. The necessity of including the Extended Kalman Filter to smooth the signal applied in the control valve is shown from a nonlinear mathematical model of a real plant.