Recent works [21]-[27] has implemented a multi-agent system
to control the operation of a microgrid. In the literature, there
are works which seek to optimize the operation of power
generation systems, using multi-layer neural networks [2]-[4].
In [2], the optimal operation is calculated by dynamic
programming method, and this information is used as the
training data for the neural network. In [3]-[4], the optimal
operation patterns are calculated by a computer program under known requirements which minimize fuel consumption.
Afterward, the neural network is trained with this optimal
operation and then it is tested with other patterns which are not
used for training.