Fault diagnosis in distribution power systems using stationary wavelet transform and artificial neural network

Short-circuit faults are the most commonly occurred transient events in a distribution system. Therefore, it is necessary to analyze fault transients to detect and localize. Detection and localization of faults in a distribution power system are very difficult due to the complex structure of the system. This paper presents an efficient time-frequency based detection and localization algorithm for distribution system faults. The proposed algorithm suggests a feature extraction from the transient signal using Stationary Wavelet Transform and machine-learning using Artificial Neural Network to detect and localize fault transients. The result obtained in this study proves the reliability of the proposed algorithm by achieving better accuracy in fault detection and localization.