A Day-ahead Centralized Unit Commitment Algorithm for A Multi-agent Smart Grid

International audienceRenewable energy resources like wind and solar have become an effective factor in the energy production on the planet as they are inexhaustible renewable resources. However, they are very intermittent and their output cannot be predicted certainly. In this paper an algorithm of unit commitment within a power grid integrating wind and photovoltaic production units is proposed in a centralized approach that takes into account provisional data about the renewable energy production. Here the unit commitment problem is stated as a power demand coverage problem with some prespecified merit order list. A multi-agent architecture is proposed to facilitate the message exchange and easy addition and deletion of agents in the grid. This architecture is flexible and easy reconfigurable as it can provide solutions under assumptions of a decentralized approach. An implementation using JADE platform is presented in this work. The system is tested using real-data sets from an existent energy transport network in France (RTE). The results based on different operating conditions show the economic sense of the proposed strategy