WAVELET TRANSFORM BASED DE-NOISE ALGORITHM BY THRESHOLDING IN ON-LINE PARTIAL DISCHARGE DETECTION

One of the key problems in on-line partial discharge (PD) detection is to suppress many of the sinusoidal signals, which interfere the PD identification. Under different scales and using orthonormal multiresolution analysis based on wavelet transform, the signals can be decomposed into many different levels. By appropriate processing of the decom-position coefficients on the scales according to the threshold values the interference signals can be effectively eliminated and the de-noised signals can be obtained. Through the simulated and on-site experiments, it is proved that this method is very effective.