Transmission line fault location analysis using deep network designer / Nurin Fatini Mohd Nizam

This study presents significant advancements in the application of Long Short-Term Memory (LSTM) networks for fault location analysis in transmission lines, a method not extensively explored in previous research. By utilizing an LSTM model to analyse fault signals from a 100 km transmission line with a voltage rating of 400 kV, the research demonstrates robust performance in accurately identifying fault locations, addressing the limitations of traditional machine learning methods that rely heavily on feature extraction and are sensitive to specific line parameters. The model's performance was benchmarked against four other fault location techniques, revealing that although the LSTM with 500 epochs exhibited lower accuracy initially, it highlights the potential for improved performance through further training. Notably, this study emphasizes the use of root mean square error (RMSE) as a metric for evaluating fault location accuracy, providing a nuanced understanding of model performance that is relatively rare in existing literature. Furthermore, the findings suggest that while LSTM models may face challenges when trained on specific transmission lines, there is substantial potential for generalization across different lines with continued refinement. Overall, this research contributes valuable insights to the field of electrical engineering and machine learning applications in power systems, paving the way for future innovations to enhance reliability and efficiency in fault detection and location.