147. EFFICIENT FEATURE EXTRACTION AND CLASSIFICATION OF POWER QUALITY DISTURBANCES
The increasing number of renewable energy sources, together with the installation of modern control equipment requires higher power quality (PQ) in the generation, transmission and distribution systems. In order to maintain and improve the power quality, power disturbances should be monitored continuously. Power quality monitoring and analysis must be able to detect and classify the disturbances present in voltage or current waveforms. In this paper a method for feature selection and classification of power quality disturbances using wavelet transform (WT) and random forest (RF) algorithm is proposed. The classification results for seven and ten PQ classes are compared with results obtained by applying some previously published methods, proposed by different authors. Moreover, a comparison of the classification accuracies obtained in noisy environment, by using our method and the other proposed wavelet based methods, is made. The investigation has shown that the proposed method represents an efficient method for feature extraction and classification with high accuracy.