Fault Detection and Identification for DC Microgrid with Wavelet-Based Artificial Neural Networks

Abstract

The widespread implementation of DC microgrids (DCMGs) is a significant step toward future power systems' ability to match load requirements with distributed generation precisely. DC microgrid has become a viable alternative for increasing DC applications and load needs compared to AC microgrid. However, the DC microgrid system's thriving potential is hampered by the significant challenges associated with its protection. The challenges arise due to the time constraints imposed by rapidly increasing fault currents in DC systems, an absence of frequency and phasor information, and the absence of a natural zero crossing of DC fault current. Furthermore, altering DC microgrid topologies impacts the existing protection mechanisms significantly. As a result, for the DC microgrid to operate adequately, an intelligent protection strategy is required. This work presents intelligent fault detection and identification approach for DC microgrids based on wavelet transform (WT) and artificial neural networks (ANNs). In this work, firstly, the wavelet transform is applied for pre-processing the current signals to determine the detailed wavelet coefficients. Then, the maximum value among the detailed coefficients, which provides critical information during fault events, is used to construct the input feature vector. After extracting the fault characteristics, the data set, consisting of input feature vector along with output tags, is utilized for training an Artificial Neural Network (ANN) model to identify and categorize faults in DCMGs. For data collection, the study simulates diverse fault types, fault locations, fault resistance, and fault incident time and nofault (load variation) scenarios under both grid-parallel and off-grid operating modes. MATLAB/Simulink software has been used to run the simulations on a PV-based DC microgrid. The proposed scheme's test analysis using ANN verifies the scheme's reliability and efficiency in providing potential DC microgrid protection.