A noncooperative distributed model predictive control for constrained linear systems with decoupled dynamics

This paper addresses the distributed MPC (model predictive control) for a set of linear local systems with decoupled dynamics and a coupled global cost function. A previous approach is improved. For each local controller, the infinite horizon control moves are parameterized as a series of free control moves followed by a single state feedback law. An interacting stability condition is derived, disassembled and incorporated into the design of each local control so as to achieve the stability of the global closed-loop system. Each local system exchanges with its neighbors the current states and the previous optimal control strategies. The global closed-loop system is shown to be exponentially stable provided that all the local optimizers are feasible at the initial time.