A DWT-RNN-Assisted Intelligent Differential Protection Scheme for Grid-Tied and Islanded DC Microgrid

Abstract

AbstractThe integration of renewable sources into a DC microgrid offers clear advantages such as high efficiency and effective simple control. Despite this, the high penetration of distributed generators can cause problems such as safety, and islands in the microgrid can affect the reliability of the power system. Because the nature of the fault current in the DC, microgrid is different from that of AC, which brings a great challenge for fault detection and diagnosis. To address this issue, this paper presents an intelligent differential protection scheme for microgrids that employ the machine learning (discrete wavelet transform and ANN) method. The proposed differential protection scheme employs DWT for fault detection and ANN for event classification. The discrete Fourier transform is used in this study to preprocess the voltage and current signals in order to calculate the microgrid fault formation. Subsequently, available ANN-based classifiers are presented to evaluate the proposed scheme’s efficiency in terms of defect detection, identification, and classification. This study collects data from various aspects by simulating different fault and no fault cases for the microgrid configuration in grid-tied and island modes of operation. The simulation is performed on a standard medium-voltage microgrid employing MATLAB/Simulink. The results also show that the proposed method can detect the faults in grid-tied and islanded microgrids.KeywordsDistributed energy resourcesDC microgridsArtificial neural networksDiscrete wavelet transform (DWT)Fault detectionIslanding detection techniquesMicrogrid protection