Research on Noise Recognition and Suppression of Partial Discharge On-line Monitoring in Transformer

概要

論文の詳細を見る
Partial discharge (PD) is one of the major factors causing insulation damage of power transformers. PD signal usually mixed up all kinds of noise, it had some limitation to extract faint PD impulse signal from powerful noise with traditional analytic methods. Based on the analysis of the noises correlation, spectral relative distribution and frequency, the noises unique characteristics were obtained. Adaptive algorithm based on wavelet packet theory was presented to suppress the periodic noise. The wavelet transform was used to reject white noise and extract characteristic parameters from PD signal. Piecewise cubic spline interpolating algorithm was presented to reconstruct the wavelet coefficients. Compared with the alternate projection method, the algorithm is effective and easy to implement, which can get higher signal-to-noise ratio gain and smaller root mean square error than the alternate projection method. The results show the method can be used for noise recognition and suppression on-line monitoring of partial discharge in transformers.
2007-11-01