Detecting winding to ground fault locations in power transformers using back-propagation neural networks

This paper presents an algorithm based on a combination of discrete wavelet transforms and neural networks for detecting locations of winding to ground faults in a two-winding three-phase transformer. The fault conditions of the transformer are simulated using ATP/EMTP in order to obtain fault current signals used as an input for a training process of a back-propagation neural network. The training process and fault diagnosis decision algorithm are implemented using toolboxes on MATLAB/Simulink. Various cases studies based on Thailand electricity transmission and distribution systems are performed to verify the validity of the algorithm. It is found that the proposed method gives a satisfactory accuracy, and will be particularly useful in a fault diagnosis process for a transformer manufacturer.