Advanced transmission line fault protection including the voltage sag.

The article focuses on a hierarchical protection framework for detecting and classifying voltage sag (VS), high impedance faults (HIF), and low impedance faults (LIF) in low-voltage transmission lines integrated with fast-charging stations (FCS) and photovoltaic (PV) systems, modeled on an IEEE-14 bus power system. The methodology employs a two-layer autoencoder for feature extraction from voltage and current signals, followed by a long short-term memory (LSTM) network for initial classification into VS, HIF, and LIF, and then uses Extreme Gradient Boosting (XGBoost) to further classify symmetrical and asymmetrical fault subcategories such as single-line-to-ground (LG), line-to-line-to-ground (LLG), three-phase (LLL), and three-phase-to-ground (LLLG) faults. The proposed approach achieved high accuracy—99% for VS detection and near-perfect precision and recall for fault classification—while demonstrating robustness to noise, computational efficiency, and scalability suitable for real-time applications. This integrated machine learning framework addresses challenges posed by the increasing penetration of FCS and renewable energy sources, enhancing the reliability and stability of modern power grids.