Proposal of enhanced classification of disturbing phenomena in electrical networks

The author proposes a method of power quality classification using SVM (Support Vector Machines) neural networks. It shoves enhanced performance in comparison to a classifier based on the RBF (Radial Basis Function) network. Both, RBF and SVM networks are introduced and described as appropriate tools for classification problems. Space phasor is used for feature extraction needed by the classifiers. Different disturbance classes has been simulated (e.g. sags, voltage fluctuations, transients) using Matlab modeling. The classifiers has been tested with these signals.