FAULT LOCATION TECHNIQUE BASED ON MAXIMUM TRASIENT DEVIATION AND LSTM NEURAL NETWORK

Resumo

In electrical power systems, transmission lines (TLs) are critical components that are often susceptible to different types of faults. Once a fault occurs, a fast restoration procedure is crucial and depends on accurately determining the fault location. This study introduces a novel method that combines Absolute Maximum Transient Deviation (AMTD) measurements and Long Short-Term Memory (LSTM) Neural Networks for fault location in TLs. The definition of the most suitable LSTM topology is discussed, and the training process is elaborated to identify the best configuration for performing the desired task. The simulated fault cases and the AMTD-based algorithm were modeled and executed using the Real-Time Digital Simulator (RTDS), enabling real-time simulations. The results demonstrate that the developed method is accurate for fault location in TLs, regardless of fault characteristics.