In this paper, a novel technique based on [32], which
employs a recurrent neural network has been proposed to
achieve the optimal operation of a microgrid interconnected to
an utility grid. This RNN is combined with a multi-agent
system to operate the global system, fulfilling the load
demand, minimizing the power supplied by the utility grid and
maximizing the one supplied by renewable sources. The
results show that the algorithm determines the optimal power
values for a time horizon of one day, for the wind, solar,
batteries systems and the utility grid. As future work, we will
make the physical implementation of the microgrid, in order to
obtain experimental results.