This research presents a comparative analysis of two artificial neural network (ANN)‐based methods, two‐end and one‐end, for fault localization, classification, and detection in power transmission lines. The methods utilize the magnitude‐squared values from the fast Fourier transform (FFT) of current signals obtained from both ends and one‐end. These values serve as input features for the ANN in both approaches. Four distinct models were employed: a detection model, a classification model, and two location models with different architectures, resulting in a total of eight models for assessing localization accuracy. The methods were evaluated via the IEEE 9‐bus system in MATLAB/Simulink under various fault scenarios with different fault types and locations. The results demonstrate satisfactory performance in fault detection and classification with high accuracy and fast response times. Notably, the two‐end location models provided more precise fault location accuracy than did the one‐end models. Overall, the proposed methods exhibit high reliability and accuracy in fault detection, classification, and localization.
Touati et al. (Thu,) studied this question.