Classification of power quality disturbances based on KF-ML-aided S-transform and multilayers feedforward neural networks

Classifying power quality (PQ) disturbances is one of the most important issues for PQ control. The S-transform (ST)-based neural networks in conjunction with Kalman filter based on maximum likelihood (KF-ML) are presented for classification of PQ disturbances. To accurately extract features in high-noise cases, the KF-ML is used to remove noise from the original distorted waveform. Then, ST technique is used to extract the significant features of disturbances. Finally, a classifier based on multilayers feedforward neural networks can accurately recognise various types of PQ disturbances. Six simulated single disturbances and six complex ones with different noise levels are tested for the sensitivity to noise. Classification results show that the classification accuracy of the proposed method is more than 95% even in 20 dB high-noise condition, and also validate the superiority of strong rejection to noises. Comparison studies between the proposed method and other classification methods are also reported to show the advantages of the proposed approach.