Advanced transmission line fault protection including the voltage sag

The rapid adoption of electric vehicles and the increasing integration of fast-charging stations (FCS) into modern power networks have introduced significant challenges to system stability and reliability. Critical issues, such as voltage sag (VS), high impedance faults (HIF), and low impedance faults (LIF), disrupt system operations and can cause severe equipment damage. This study proposes a novel hierarchical protection framework tailored for an IEEE-14 bus power system with eight concurrently operating FCS supported by photovoltaic systems and the main grid. The framework employs a hybrid machine learning approach: a two-layer autoencoder for feature extraction, long short-term memory networks for initial fault categorization into VS, HIF, and LIF, and XGBoost for refining fault classification into subcategories, including single-line-to-ground, line-to-line-to-ground, three-phase, and three-phase-to-ground faults. The proposed methodology achieved 99% accuracy in VS detection and near-perfect precision and recall for fault classification, significantly outperforming existing techniques. Furthermore, the framework offers enhanced computational efficiency, scalability, and robustness to noise, making it a reliable solution for real-time fault detection and classification. This work provides a foundation for advancing intelligent fault management systems in modern power grids with high FCS penetration, addressing both current and future challenges.