A distributed hierarchical architecture for community-based power balancing

The coordination and control of the agents in electrical networks is expected to have an important positive impact in future smart-grids. In the present paper we propose a distributed hierarchical coordination framework for power balancing, and the algorithms required to build such coordination architecture. This architecture is built such that a group of agents (a subtree of the hierarchy) can respond and adjust to deviations from a day-ahead plan for that group, without the need to coordinate with its upper levels of the hierarchy at every time step. The agents are arranged in groups by taking into account a measure of the flexibility of each group, thus allowing a partial de-coupling of the groups during the coordination, which reduces the required work of the central coordinator and gives robustness to the system. The assignment of agents to groups is formulated as a welfare maximization problem, a NP-hard problem for which an approximate solution can be obtained efficiently by using the sub-modularity and monotonicity of the involved functions. An electric vehicle charging coordination scenario is presented to exemplify the proposed architecture and methods.