Multi-agent systems for grid energy management: A short review

The electric grid is currently undergoing important changes: it is evolving from an entirely centralized structure to a decentralized one, mainly due to the massive development of distributed renewable energy sources. This evolution toward what we now call the smart grid requires new control methods. These methods must be able to withstand new requirements, such as the highly distributed nature of the grid, the ability to run in islanding mode, the intermittency of renewable energy sources and the limited bandwidth for communications. Multi-agent systems (MAS) have characteristics that meet these requirements. In contrary to classical analytical methods, the grid is considered as a collection of simple entities called agents corresponding to sources, loads and other components, evolving in a given environment. A certain degree of distributed or collective intelligence can be achieved through the interaction of these agents with each other, cooperating or competing to reach their goals. This article is a short review of several multi-agent systems published in the literature and used for grid energy management. It presents a number of concepts and experiments used by researchers to apply this promising method. Various approaches and their results are compared, in order to give the reader a global perspective of the state of the art in this precise domain.