Nonlinear model predictive control using a Wiener model of a continuous methyl methacrylate polymerization reactor

A subspace-based identification method of the Wiener model, consisting of a state-space linear dynamic block and a polynomial static nonlinearity at the output, is used to retrieve the accurate information about the nonlinear dynamics of a polymerization reactor from the input-output data. The Wiener model may be incorporated into model predictive control (MPC) schemes in a unique way that effectively removes the nonlinearity from the control problem, preserving many of the favorable properties of the linear MPC. The control performance is evaluated by simulation studies, for which the original first-principles model for a continuous methyl methacrylate polymerization reactor takes the role of the plant while the identified Wiener model is used for control purposes. On the basis of the simulation results, it is demonstrated that, under the presence of strong nonlinearities, the Wiener model predictive controller (WMPC) performed quite satisfactorily for the control of polymer qualities in a continuous polymerization reactor. The WMPC strategy proposed is validated by conducting an online digital control experiment with an online densitometer and viscometer. It is observed that the WMPC performs satisfactorily for the polymer property control of the highly nonlinear multiple-input multiple-output system with input constraints.