Hybrid Ensemble Learning Technique for Efficient Fault Detection in AC Microgrid

Microgrids are localized energy systems integrating distributed energy resources (DERs) such as solar, wind, and conventional generators to enhance grid reliability and sustainability. However, their decentralized nature and bidirectional power flow introduce significant fault detection challenges. Traditional protection schemes struggle with dynamic fault conditions, necessitating advanced machine learning (ML) techniques for improved accuracy. This paper presents an ensemble learning approach incorporating a Random Forest Classifier, XGBoost, and Support Vector Classifier (SVC) to detect faults in a microgrid environment. The proposed model leverages historical voltage and current data to classify different fault types, achieving a high detection accuracy of 99.2 %. Simulation results demonstrate the model's effectiveness, outperforming conventional methods and individual classifiers in fault identification.