Distributed model predictive control via decomposition-coordination techniques and the use of an augmented Lagrangian
Résumé
This paper is devoted to distributed nonlinear predictive control design through the use of both an augmented Lagrangian formulation and price - decomposition - coordination. We show how Lagrangian relaxation can be used to design a distributed MPC scheme, which allows dramatic reduction of the computational requirements for solving large-scale nonlinear MPC problems due to computation parallelism. The effectiveness of this approach is demonstrated for the so-called Load Frequency Control of power systems.