Effective Machine Learning Models for Fault Detection and Classification in Power Transmission Lines

Effective fault detection and classification in transmission lines is crucial for maintaining system stability and security. Undetected or misclassified faults can lead to power outages, equipment damage, and significant financial losses. This study investigates the application of machine learning (ML) models for this purpose. A transmission line model comprising of generators, RLC loads, and transformers is developed using MATLAB Simulink, and used for generating two datasets: one for fault detection and the other for fault classification. A wide range of ML models are evaluated to determine their effectiveness for fault detection and classification tasks. A comparative evaluation of the various ML models shows that Cat Boost ensemble model performs best for the fault detection task, achieving an extremely high accuracy of 99.42%. For the fault classification problem, the Stacking Classifier obtains the best performance, achieving accuracy of 94.3%. The results show the effectiveness of machine learning approaches for the electric fault detection and classification task. In the interests of transparency and supporting further research in this area, the Simulink model and python code of the ML models developed as part of this work are made openly available online.