MACHINE LEARNING-BASED FAULT DETECTION IN THREE PHASE TRANSMISSION LINES AND ELECTRIC MACHINES

Abstract

This research investigates the potential of machine learning algorithms to detect and categorize faults in electrical machines and transmission lines. The objective of this project is to utilize a machine learning algorithm that can accurately identify and classify faults in these electrical systems, thereby decreasing downtime and increasing overall system reliability. Further, it involves simulating fault scenarios in a MATLAB Simulink environment to generate datasets, which are then preprocessed and split into training, validation, and testing sets. The Decision trees, XGBoost, k-NN, and random forest Algorithms are employed. Additionally, Simulink is integrated with the best-performing machine-learning algorithm for real-time fault detection. The experimental results suggest that K-Nearest Neighbours (KNN) and Random Forest algorithms outperformed other tested techniques in terms of fault detection and classification in transmission lines and electrical machines.