Use of Wavelet and Neural Network (BPFN) for Transformer Fault Diagnosis

This paper investigates the application of wavelet analysis technique to transformer fault diagnosis using artificial neural network. Wavelets provide an efficient means of decomposing voltage and current signals to a detectable and discriminate features as it convolutes into different frequency components. It is being found that neural network is the most suitable tool for fault identification as it can recognize the hidden relationship between the fault status and some symptoms and predict the fault of a new sample based on previous knowledge. For the purpose of fault signal acquisition like winding-to-winding, winding-to-ground, disc-to-disc, turn-to-turn a 61mva, 11.5/230 kV transformer is used.