An Intelligent Fault Classification Method for Microgrids Based on Discrete Orthonormal S-transform and Ensemble Classifier

This paper proposes an intelligent fault classification method for microgrids using a combination of discrete orthonormal Stockwell transform (DOST) and ensemble classifier. The proposed method first pre-processes the one cycle post-fault voltage and current signals using DOST. Then, the standard deviation of the DOST coefficients of the voltage and current signals is utilized to form the input feature vector to train and test the ensemble of classifiers. The ensemble classifier created by combining the hybrid base classifiers such as k-nearest neighbor (k-NN), support vector machine (SVM), and decision tree (DT) is used for classifying faults in microgrids. To evaluate the performance of the proposed method, we conduct a comprehensive evaluation study on a standard IEC microgrid test system using MATLAB/Simulink software. The test results confirm that the proposed method classifies faults in a microgrid with high classification accuracy.