Accurate Fault Localization in Transmission Lines using Feed-Forward Neural Network and Post-Processing Algorithm

Transient stability of power system depends on very fast isolation of large disturbances such as transmission line faults. In a transmission network, when the relay detects an abnormal signal, the circuit breaker isolates the faulted transmission line from the rest of the healthy system and the fault is cleared. Traditional methods for fault localization use manual time-consuming approach. This paper presents a method for localizing faults in transmission lines using deep neural network which can instantly detect the faulted area with high accuracy. The initial prediction of the location is done through the use of artificial neural network(ANN). This ensures a fast response to a fault and enables real time fault localization. A post processing algorithm is used in conjunction with the ANN model to increase the model’s accuracy even further and find the exact fault location.