An off-line formulation of min-max MPC for nonlinear systems via multiple LPV embeddings

An off-line min-max MPC strategy is proposed for constrained nonlinear systems, whose trajectories can be embedded within those of a bank of linear parameter varying models via an embedding scheme. The off-line min-max MPC algorithm is used to determine the control law for each LPV embedding model corresponding to invariant ellipsoidal sets constructed off-line. The control law which will be implemented to the original nonlinear system is chosen according to the set in which the current state lies. This strategy can greatly reduce the online computational burden compared with the traditional nonlinear MPC.