System identification and MPC based on the volterra-laguerre model for improvement of the laminator systems performance

A Model Predictive Control (MPC) scheme in order to reduce the great electrical demand and mechanical stresses on laminator systems, which are presented at the transient states when the machine is started up under load, is reported. This study is presented in two parts: first, the system is identified through Volterra-Laguerre (V-L) model. Here, the system parameters are extracted using a set of input-output data obtained by applying a Pseudo Random Multi Sequence (PRMS) on system under study. The methodology for system identification is based in the projection of the Volterra kernels onto a Hilbert space, built by Laguerre polynomials, as a set of Orthogonal Basis Functions (OBF). Second, when the V-L model is known it enters inside a MPC for a precise control of the rolling mill speed, which is a crucial indicator in order to guarantee an optimal manufacture of vinyl tile. The MPC used in this study is inspired from the Dynamic Matrix Control (DMC) algorithm, which improves the precision on the tracking trajectory control. Simulations have shown that a MPC based on the V-L model, could give a quadratic error about 0.015%, after transient state, indicating its robustness.