Automatic classification of power quality events and disturbances using wavelet transform and support vector machines

In this study, a new approach for the classification of power quality events is presented. Also, power quality disturbances, which occur in each phase of the power system after a fault event, are classified with the proposed system. In the proposed recognition system, three-phase voltage signals are used in order to identify the type of power quality events. Three-phase voltage signals are subjected to normalisation and segmentation processes. A wavelet transform method is used in order to obtain the distinctive features of event signals. An efficient feature vector, which represents the distinctive characteristics of three-phase event voltage signals and reduces data size, is extracted by applying the two-stage feature extraction process. Power quality event types are determined by using a support vector machine classifier. At the last stage of intelligent recognition system, types of power quality disturbances regarding each fault event are identified by doing a further analysis. Real power system data are used to evaluate the performance of the proposed approach. According to the obtained results, proposed intelligent recognition system classifies power quality event types with a high accuracy. The analyses and results also show that the proposed approach is efficient, reliable and applicable.