Identification and Model Predictive Control Design of a Polymer Extrusion Process

Abstract This paper deals with the challenging problem of closed-loop identification and Model Predictive Control (MPC) design in multivariable chemical processes and particularly in a co-rotating twin screw extruder processing powder coatings. To this aim, identification tests based on step response excitation signals were designed to estimate a low order process model in order to assist the model based control design of the process. Then, a discrete time predictive controller was developed to regulate the extruder and improve its control performance and disturbance rejection properties. The predictive control strategy exploits sets of Laguerre orthogonal functions to express the projected control trajectory. The process under MPC control was tested using different simulation scenarios and the results have shown very good performance with fast settling times and good disturbance rejection properties.