A robust intelligent model for identification of power quality signals using deep learning and feature fusion algorithm in smart grid

This paper presents a novel approach to identify powers quality (PQ) signals using short-time Fourier Transform (STFT), deep learning and feature fusion algorithm. STFT is well-known signal processing technique that efficiently analysis PQ signals with respect to time-frequency domain. The STFT captures spectrogram of signal and complex PQ signals and fed the input to proposed one-dimensional convolutional neural network (1D-CNN), the proposed deep learning approach consists two parallel CNNs and a bi-directional long short-term memory (Bi-LSTM) unit. The proposed BiLSTM network is essential to improve feature learning ability as compared to traditional methods. In order to validate the proposed intelligent model to identify sixteen classes of single and complex PQ signals in PV-SOFC based smart grid networks, a Simulink model with photo-voltaic (PV) and solid-oxide fuel oxide (SOFC) is built on PSCAD/EMTDC software. Further, noisy PQ signal data is tested with real-time PQ signals collected from online databased of power quality monitoring.