On-line recognition and retrieval of PD signal by regularity measurement based on computation of Lipschitz exponents in wavelet domain

The problem of on-line recognition and retrieval of relatively weak industrial signals such as partial discharges (PD), buried in excessive noise, has been addressed in this paper. The major bottleneck being the recognition and suppression of stochastic pulsive interference (PI), due to the overlapping broad band frequency spectrum of PI and PD pulses. Therefore, on-line, on-site PD measurement is hardly possible in conventional frequency based DSP techniques. In the authors methodology, the observed noisy signal is enhanced and non-pulsive noises are removed by implementing a wavelet based soft-thresholding scheme. Then, the pulses are detected using a simple peak-detector and further analysis is done on localized pulses by taking an appropriate window at the detected location. The features of the PD and PI pulses are obtained by measuring the regularity of the pulses at the detected location, which is accomplished by estimating Lipschitz exponents (LE). In this regard, a wavelet based methodology has been used in estimating LE, which is then used as an index for classifying the pulses into PD and PI. The proposed method was found to be effective in automatic retrieval of PD pulses.