A Single-End Line-to-Ground Fault Location Method for HVAC Transmission Lines Using Sequential Feed-Forward Neural Networks

The paper presents an Artificial Neural Network (ANN)-based single-ended fault location technique as a practical and robust solution for improving the protection and maintenance of High Voltage Alternating Current (HVAC) transmission systems. The proposed fault location method depends on utilizing multistage predictions implemented by sequential feed-forward neural networks. The proposed method is a single-end-based method that does not necessitate the acquisition of measured signals from the other end. The inputs to the first stage are the voltage and current sequence components derived from the measured signals at the sending end of the transmission system. The training data are optimized using fault analysis theorems to improve the accuracy of fault location prediction. The fault point voltage, obtained from the first stage, is identified as a key parameter. The rationale and justification for its effectiveness and significance are presented in the paper. This parameter is subsequently combined with selected primary data to serve as the input for fault location prediction in the final stage. Extensive testing is performed with the assistance of MATLAB software, considering the effect of various fault scenarios, including different fault resistances, fault inception angles, and load angles. The findings indicate that the maximum error of the proposed method does not exceed 1%, confirming its efficacy in locating line-to-ground faults in HVAC transmission lines.