Classification of power quality disturbances using Wavelet and Artificial Neural Networks

An automated classification system based on Wavelet transform as a feature extraction tool in combination with Artificial Neural Network as algorithm classifier is presented. Perturbed signals generated according to mathematical models have been used to obtain experimental results in two stages, first, with a data set with simple disturbances and, later, including complex disturbances, more usual in real electrical system. In both cases noise is added to the signals from 40dB to 20dB. Two different neural networks have been used as classifier algorithm, a backpropagation and probabilistic. A data set with several disturbances, simple and complex, has been generated by simulation software based on electrical models, to test the implemented system. Evaluation results verifying the accuracy of the proposed method are presented.