Polymerization reactor control using autoregressive-plus Volterra-based MPC

A nonlinear model-predictive control scheme based on the autoregressive-plus Volterra model is proposed. A ''plant-friendly,'' yet persistently exciting, input sequence is designed for model identification. A second feature of this work is that the control action is determined by solving a less computationally burdensome nonlinear programming problem than the optimization problems associated with Newton-type controllers and polynomial ARMA model nonlinear MPC. A third contribution is that semiglobal closed-loop stability conditions are derived and are shown to be less conservative than those previously published. The control scheme is shown to outperform a feedback strategy using proportional integral control and a linear model-predictive control design for the control of two polymerization reactor case studies. The complex features of the latter case study [multivariable (3 x 3), recycle loop, high-order (24) dynamics] motivate the applicability of this approach for industrial problems.