Pattern recognition of phenomena associated to power quality using neural networks

This paper presents an approximation to the recognition of phenomena associated to power quality in electric networks, by means of neural networks and based on the work discussed in reference [Taboada, JD, et al., (2003)], where a former recognition of the phenomena was processed using the discrete wavelet transform (DWT). A multiple layer perceptron (MLP) was used, together with the back propagation algorithm for the training process. The patterns recognized corresponded to signals of harmonic, transient, sags and swell waveforms.