LSTM-based low-impedance fault and high-impedance fault detection and classification

In this article, a long short-term memory based protection scheme for power transmission lines is presented. A fault detection framework is developed that uses voltage and current signals' RMS values as input. The proposed work is established for various shunt faults, both low-impedance and high-impedance faults which are tested on a standard IEEE 14 bus system and an existing real transmission network. Results confirm the detection and classification of faults with accuracy and precision higher than 99%. The impact of non-faulty events such as load switching, capacitor switching, noisy data, and load variation conditions is also studied to analyze the model's performance. The efficacy of the proposed method is confirmed by comparing it with different methods in the literature. Results indicate the aptness of the proposed scheme for the protection of power transmission lines.