High Sensitive Distinction of Discharge in Air by Daubechies Wavelet Transform

If partial discharge occurs in high voltage apparatus, it is unfavorable in view point of its insulation reliability, because they might develop into its insulation degradation or its electrical breakdown. In order to raise the insulation reliability of an apparatus, it is important to detect a minute partial discharge with sufficient sensitivity, especially suppressing background noise. This paper deals with the waveform processing technology by the Daubechies wavelet transform to make relief of the partial discharge signal from a measured noise-containing signal. On this basic idea, here is discussed that the optimal Daubechies order and its level have a close relation with the detection impedance and the sampling interval of the measured signal. Since the partial discharge waveform measured with the detection impedance of parallel circuits of RLC tuned into a damped oscillatory pulse, it has been demonstrated that the Daubechies wavelet transform is effective in discriminating the partial discharge signal from the measured noise-containing signal. Moreover, by choosing suitably the Daubechies order and its level applied to the measured data, it has been clarified that even a minute glow corona which have been masked by the background noise, and also the streamer corona turns into clear appearance on the transformed wave with sufficient sensitivity.