Application of artificial neural networks for discrimination of nonlinear loads

Smart Grids requires a complex infrastructure. It is associated with the advanced metering system which is responsible for data acquisition and processing. Taking into account this new paradigm, this paper proposes the application of Artificial Neural Networks for classification of nonlinear loads that are connected to the electrical system by using signals processed for each sample. For this purpose, multilayer perceptron and radial basis functions neural network have been used for training and validation. The results have shown good results such as the performance above 90% of accuracy in the correct classification.