Constraint-adaptive MPC for large-scale systems: Satisfying state constraints without imposing them

Abstract Model Predictive Control (MPC) is a successful control methodology, which applied to increasingly complex systems. However, real-time feasibility of MPC can be challenging for systems, certainly when an (extremely) large number constraints have adhered to. For such scenarios with state constraints, this paper proposes two novel schemes general nonlinear we call constraint-adaptive MPC. These dynamically select at each time step (varying) set that are included in the on-line optimization problem. Carefully selecting significantly reduce, as will demonstrate, computational complexity often only slight impact on closed-loop performance. Although not all (state) imposed optimization, still guarantee recursive and constraint satisfaction. A numerical case study illustrates proposed demonstrates achieved computation improvements exceeding orders magnitude without loss