Power quality event detection and analysis in power system using wavelet transformation

By means of wavelet transform and neural network, a novel approach for power quality (PQ) event classification is proposed. The feature vectors for various PQ events are extracted using wavelet transform which can accurately localizes the characteristics of signal in time-frequency domains. The feature vectors are then applied to the neural network for training and PQ pattern classification. The neural network have demonstrated promise in pattern recognition and been considered a potential alternative approach to pattern recognition It is concluded that the proposed neural network has better data driven learning and local interconnections performance by comparing with a classic neural network. The comparison between the proposed method and the other existing method is discussed. It is proved that the proposed approach can provide accurate classification results and give a new way for detection and analysis of power quality events.