Open-loop and closed-loop optimality in interpolation MPC (Q2776231)

Model predictive control (MPC) algorithms for a class of input affine nonlinear systems are analysed and compared. Predicted future control trajectories are obtained by interpolating between an unconstrained optimal control law and a control law which locally stabilizes the constrained plant. Two techniques are discussed: addition of a terminal weight in the receding horizon objective, and inclusion of an explicit convergence constraint on closed-loop trajectories. The open-loop analysis is performed in order to reveal the relationship between these approaches. It is shown that the convergence constraint approach gives a smaller open-loop predicted cost and also achieves better closed-loop performance over the more conventional terminal penalty approach. The results of numerical simulation of a second-order single-input bilinear system with nonminimum-phase characteristics are provided.NEWLINENEWLINEFor the entire collection see [Zbl 0976.00019].