Syncnet network application for identification of electrical charges

Resumo

The search for optimization of energy resources is growing every day, whether for environmental,
financial or economic reasons. In industries and homes, knowing which equipment is turned on, when it is turned
on, how often it occurs and how long it is used is essential information for energy optimization algorithms.
Therefore, recognizing the characteristic signature of the equipment is a big step. This work proposes the use of a
convolutional network using a Sincnet layer initially proposed by Mirco Ravanelli and Yoshua Bengio, used for
human speech recognition, the dataset used is the WHITED which has the signature of several loads of different
devices, this work will Evidencing the tests carried out on isolated loads and their identification, the success of this
work will allow this network to carry out tests of coupled loads in the future and, in the future, integrated to an
energy optimization system, monitoring a single point in the electrical network.