Detection and classification of power quality disturbances using STFT and deep neural Network.

Distribution networks with renewable energy sources (RESs) are rapidly integrated to meet the energy demands, and build for hybrid power systems. This causes the power quality (PQ) problems due to use of non-linear loads in the distribution networks. This paper proposes an effective deep leaning (DL) architecture with signal processing technique to detect and classify single and composite types of power quality disturbances (PQDs). Main features of PQDs has been extracted from the signals by using Short-time Fourier transform (STFT) and fed into 2D- convolutional neural network to classify data automatically. This technique has the advantage to skip the manual feature selection of PQDs events. The proposed work is compared with other advanced type of deep neural networks (DNNs) to prove the effectiveness of the classifier. Moreover, IEEE-13 node system is simulated in MATLAB/Simulink to create PQDs samples and validate the performance of the proposed method. The classification results show that the new STFT based DNN method is suitable to classify single and composite PQDs.