Long horizon model predictive control for nonlinear industrial processes

This paper considers an MPC algorithm for nonlinear plants. The MPC problem amounts to solving aquadratic programming problem. A structured interior-point method (IPM) is proposed to solve the MPC optimization problem so as to obtain a feasible computational effort for longer prediction horizons. We compare the proposed method with standard QP solvers. The choice of the QP optimizer is investigated for two nonlinear industrial cases, namely an evaporation process and a high-purity distillation column. The effectiveness of the structured IPM is demonstrated for long horizon MPC controller design of the first process.