Fault classification and localization in power transmission line based on machine learning and combined CNN-LSTM models
Nguyen Quoc Minh, Nguyen Trong Khiem, Vu Hoai Giang
Abstract
Accurate classification and localization of faults in power transmission lines are critical for maintaining grid stability and minimizing downtime. However, it remains challenging due to the variability of fault conditions and system complexities. In this paper, we present a novel method to classify and localize faults in power transmission line using machine learning algorithms. The IEEE 9 Bus is simulated in MATLAB Simulink and faults are generated with various conditions including fault types, load level, fault resistance, and fault location. More than 300 thousand fault datasets are created for both classifying and localizing tasks. Support Vector Machine (SVM), Decision Tree (DT), Logistic Regression (LR), XGBoost, and Artificial Neural Network (ANN) are used for fault classification while Convolutional Neural Network (CNN), Long-Short Term Memory (LSTM) and combined CNN-LSTM models are used for fault localization. The results show that XGBoost can classify fault with the accuracy of 99.82 %, while the CNN-LSTM can localize fault with MAPE < 1 % and MAE < 0,16 km.