Coalitional model predictive control: Optimization, heuristics, and robustness

Abstract

This doctoral thesis focuses on coalitional techniques for controllers based on model predictive control (MPC). These controllers are considered agents that form coalitions to collaborate in the control of the overall system, taking into account the interactions between subsystems and varying operating conditions while ensuring that constraints are satisfied, feasibility is maintained, and stability is preserved. First, we present the challenges of large-scale systems, review existing methods in the literature for control, introduce the main objectives of this research, and also list the articles that compose this thesis. Second, our contributions and results obtained in the articles are presented, which can be classified into three main topics: i) the optimization of cooperation topology in hierarchical coalitional MPC schemes, ii) the use of light and heuristic mechanisms for coalitional control to avoid the combinatorial explosion of control topologies when dealing with numerous agents, and iii) a robust strategy for ensuring recursive feasibility and stability despite variations in disturbances caused by changing control topologies and plug-and-play operations. We also show the manuscripts of the published and accepted articles that compose this dissertation. Finally, the main findings of this thesis and possible future research directions are provided.