Neural classification of power quality disturbances: An application of the wavelet transform and principal component analysis

This paper proposes a different method of power quality disturbance classification combining discrete wavelet transform (DWT), principal component analysis (PCA) and neural networks. This method associates properties from the multiresolution-analysis (MRA) technique with standard deviation and average calculation to extract the discriminating features from distorted signals at different resolution levels. Subsequently, a PCA algorithm is used to reduce the feature space dimension by mapping the obtained feature set into a set of fewer independent elements. Then, a radial basis function network (RBF) is employed to perform the classification of disturbances. In order to evaluate the proposed method, classifications with and without the PCA algorithm are performed.