Hierarchical distributed ADMM for predictive control with applications in power networks

In this paper, we investigate optimal control and operation of a network of linear, physically decoupled systems with a coupling in the objective function. To deal with the corresponding distributed control problem, we propose a new Model Predictive Control (MPC) scheme based on the Alternating Direction Method of Multipliers (ADMM). In particular, we thoroughly investigate the flexibility of the proposed hierarchical distributed MPC algorithm with respect to both its plug-and-play capability and changes in the (local) system dynamics and objective functions at runtime. Moreover, we show linear scalability in the number of subsystems. The efficacy of the distributed optimization algorithm embedded in MPC is illustrated on three battery scheduling problems arising from the predictive control of residential microgrid electricity networks.