(101c) Controlling Large-Scale Systems with Distributed Model Predictive Control

This talk presents an introduction and overview of recent developments in distributed model predictive control (MPC). Distributed MPC may be used to control large-scale systems in which it is undesirable or impossible to control the entire system with a single, centralized control system. The spectrum of approaches including decentralized, distributed (noncooperative and cooperative), and centralized are briefly compared with respect to the complexity of the online optimization problems and the level of required online communication. Cooperative, distributed MPC emerges from this comparison as one desirable alternative having both low computational cost and strong theoretical properties, such as robust stability and good disturbance rejection. Several current open challenges in both the theory and industrial application of cooperative, distributed MPC are briefly presented.