Analysis of Intelligent Machine Learning Techniques for the Protection of AC Microgrid

Abstract

The rapid integration of renewable energy sources into the power grid has necessitated the development of intricate protection techniques to ensure the stability and reliability of AC microgrids. This research paper examines the advancement and comparative assessment of intelligent machine learning (ML) based protection strategies for AC microgrids. Five well-known machine learning models, namely Random Forests, Support Vector Machines (SVM), Gradient Boosting, Logistic Regression, and K-Nearest Neighbours (K-NN) are assessed for their effectiveness in predicting important parameters such as voltage, current, and power in different energy components. These components include batteries, grids, photovoltaic (PV) systems, solid oxide fuel cells (SOFCs), gearbox systems, and wind energy systems. The research utilizes performance criteria, including accuracy, precision, recall, F1-score, macro average, and weighted average, to determine the most efficient models for improving microgrid safety. The results emphasize that the K-NN model is the most resilient, with Gradient Boosting and Random Forest models following closely behind. On the other hand, SVM and Logistic Regression models demonstrate poorer performance, indicating their limited usefulness in intricate energy systems.