Arquitecturas Centralizadas de Coordinación. Extrapolación del Fútbol de Robots al Control de Modo de Operaciones de Micro-Redes

Abstract

[EN] Coordination architectures in multi-agent systems represent an important topic, due to the agents must be organized in order to get a specific objective. In these kinds of systems, robot soccer is shown as an environment with a common frame for research in team coordination. The importance of these kinds of architectures is that the application in other kinds of systems which require coordination. An islanded micro-grid corresponds to a system which can be modeled as multi-agent system, where the distributed energy resources are modeled as agents are coordinated in order to protect the batteries of the energy storage systems, maintaining the common bus regulation. One way to do this throughout coordination strategies, which allow assigning roles and behaviors to the agents, depending on the system conditions. Likewise, the agents must show collaborative behaviors such as making passes in robot soccer, or the equalization of the state of charge for distributed energy storage systems, in order to obtain better results. Multi-agent systems such as robot soccer, that show uncertainties in a hostile environment, require team coordination in order to reach a common goal. One way to achieve this goal is by coordination strategies, which allow role assignment and behavior selection to the players depending on the game conditions. These strategies can present cooperative behaviors among players such as ball passes, in order to obtain better scores against an opponent team. One alternative is the design of hierarchical coordination strategies for an adequate tactic selection, roles assignment and subsequently behavior selection to the players. To this end, one possibility is to use Hierarchical Finite States Machines, due to those allow more flexible and adaptable programming, with a possibility to make changes or modifications. This allows the design of team strategies by an intuitive way according to the control architecture and the rules and specifications of the league where the team plays, in turn performing collaborative behaviors among the players. In this doctoral dissertation are proposed coordination strategies to robot soccer teams, based on tactics, dynamical role assignment and behaviors selection through the use of Finite States Machines, oriented to team architectures with centralized control and global perception. These strategies must allow an intuitive design depending on the team specifications and the rules of the league. These strategies must allow flexibility and adaptability to the necessary changes, such as the number of players, dimensions of the playing field or other changes in the league`s rules. The strategies must be show an adequate response to the failures and allow cooperative behaviors among their players. Lastly, this dissertation presents a structured architecture based on tactics, roles and behaviors for the coordinated operation of islanded microgrids. This grid architecture takes the form of a multi-agent system, and the energy storage and generation units are the team agents. The…