Abstract: As the vibration signals of the transformer body contain abundant fault information, using vibration method for on-line monitoring of transformer operation has become one of the research emphasis of transformer fault diagnosis. Based on wavelet transform and information entropy theory, the wavelet energy spectrum entropy of the transformer vibration signals are selected as characteristic signal for fault diagnosis. Using TMS320F2812 as core DSP chip, an acquisition system of wavelet energy spectrum entropy of transformer vibration signal is designed and developed. The results show that the wavelet packet energy spectrum entropy between normal operation and fault operation are different. Therefore, this method can effectively extract transformer vibration signal characteristic value and provide powerful basis for subsequent fault diagnosis.

Key words: transformer, fault diagnosis, vibration signal method, wavelet transform, information entropy, wavelet energy spectrum entropy, DSP