Robust fault detection and classification in power transmission lines via ensemble machine learning models

Area of Science:

• Electrical Engineering
• Power Systems

Background:

• Transmission lines are critical for electricity delivery but susceptible to faults.
• Faults disrupt power supply and create safety hazards.

Purpose of the Study:

• To develop a novel approach for accurate fault detection and classification in transmission lines.
• To enhance the robustness and reliability of power grids.

Main Methods:

• Analysis of voltage and current patterns across transmission line phases.
• Evaluation of machine learning algorithms: Random Forest (RF), K-Nearest Neighbors (KNN), and Long Short-Term Memory (LSTM) networks.
• Proposal of an ensemble methodology: RF-LSTM Tuned KNN.

Main Results:

• RF-LSTM Tuned KNN achieved 99.96% accuracy in multi-label fault classification.
• KNN and RF demonstrated high accuracy in binary classification (99.85% and 99.72%, respectively).
• The proposed ensemble method outperformed individual algorithms in detection accuracy and robustness.

Conclusions:

• The RF-LSTM Tuned KNN methodology offers significant advancements in transmission line fault detection.
• This approach provides valuable insights for enhancing grid reliability and stability.
• The findings contribute to ensuring a more resilient and secure power supply.