Fault Analysis in Power Transmission Systems Using Machine Learning

• Published in: 2024 IEEE Third International Conference on Power Electronics, Intelligent Control and Energy Systems (ICPEICES) pp. 494 - 497
• Main Authors: Garg, Manik, Raj, Pranav, Piyush, Prakash, Prem
• Format: Conference Proceeding
• Language: English
• Published: IEEE 26-04-2024
• Subjects: Accuracy; Classification algorithms; Decision trees; Electrical fault detection; Fault detection; Fault diagnosis; Machine learning algorithms; Measurement; Random forests; Support vector machines
• DOI: 10.1109/ICPEICES62430.2024.10719129

With the increasing complexity and decentralization of power generation and consumption, the demand for robust fault detection and classification systems in power distri-bution networks has become paramount. This project addresses this challenge by employing machine learning techniques for the detection and classification of faults in the power distribution network. For fault detection, three models, namely Decision Tree, Sup-port Vector Machine (SVM), and Random Forest, are utilized to analyze electronic metrics and data. These models prove effective in identifying and localizing faults within the network, providing a foundation for proactive maintenance. In the classification phase, Decision Tree and SVM models are employed to further categorize detected faults. The integration of these models enhances the system's ability to not only detect faults but also classify them. The results indicate that the Random Forest model demonstrates exceptionally high accuracy in fault detection, outperforming other models. In the classification phase, the Decision Tree model exhibits superior performance, providing precise categorization of detected faults. This research delves into innovative methods for predicting network faults, with a particular focus on overhead lines. The proposed approach, backed by the success of the Random Forest and Decision Tree models, demonstrates high accuracy rates using various classifiers, contributing to the development of a comprehensive fault management system.