The effectiveness of wavelet based features on power quality disturbances classification in noisy environment

Power quality (PQ) disturbances classification plays an essential role in ensuring high quality power supply of the power grid. One of the main issues in classification is how to extract the “right” features from massive amount of PQ data. The feature selection should be performed for the aim of not only increasing the classification accuracy, but in the same time reducing the calculation time of the classification algorithm. Accordingly, in this work we investigate the effectiveness of the wavelet based features on the classification accuracy in order to perform optimal feature extraction method. The investigation is made using three different classifiers, in case of pure PQ signals and PQ signals accompanied with white Gaussian noise. The results show that the effectiveness of a given feature is not general, but it depends on the kind of the other features it is used with and the noise level present in the signal.