Robust-PCA deep learning for PQ disturbances classification using Synchrosqueezing Wavelet Transform

In this paper a Robust-PCA Deep Learning algorithm using Synchrosqueezing Wavelet Transform is proposed for PQ disturbances mutli-classificalion. The algorithm was implemented and programmed in MATLAB using custom code. This approach avoids white noise, outliers and overfilling phenomena. The Synchrosqucc/ing Wavelet Transform is performed and a Robust-PCA mapping is done. External data is necessary to perform the pretreatmeni for autoscaling. A Deep Feed Forward Neural Network is implemented with 5 layers. 3 of them arc hidden layers with more than 1 million parameters to fit. The quality of the solution is validated by the cross validation of parameters, R2 and Q2. Moreover, mean square error (MSF.), ihc root of the mean square error (RMS!:), the mean absolute percentage error (M APR). Akaike information criterion (A1C) and the Schwarz information criterion (SBC) arc estimated. The adjusted R2 value is 0.989 and the RMSE obtained is 1.789. The value of R2 is 0.995. All these parameters arc calculated over the lest sel. © 2021, European Association for the Development of Renewable Energy, Environment and Power Quality (EA4EPQ). All rights reserved.