Model predictive coordinating control of dynamically interacting systems with stochastic disturbances

A computationally-efficient, model-predictive-control (MPC)-based coordinated control scheme is explored for a large-scale system with dynamically coupled subsystems. Considering a class of coordinated control problems involving consensus, the proposed formulation presents the control problem as a distributed optimization problem with sets of local and global decision variables which can be optimized online. The formulation comprises dynamic-policy-based local MPC laws whose dynamics together with the decoupled feasible sets for the subsystem decision variables are computed off-line in order to simplify the on-line computation of the local MPC moves. The overall distributed MPC optimization problem is solved using a dual-averaging-based iterative algorithm in real time. The performance of the proposed scheme is illustrated with a simulation example.