Feature extraction for partial discharge based on cross-wavelet transform and correlation coefficient matrix

A new method for partial discharge feature extraction based on cross-wavelet and correlation coefficient matrix is proposed aimed at the characteristic quantities' high-dimension and high-sensitive to noise. Cross-wavelet transform method gives analysis between time and frequency domain for signals. The method is immune to the noise and it is applied for power transformer partial discharge signal analysis. New characteristic quantities are obtained representing cross- wavelet spectrum. And then correlation coefficient matrix is constructed for correlation analysis of extracted characteristic quantities, and those which have similar classification ability are eliminated. Finally probabilistic neural networks and back propagation neural network classifiers are utilized for pattern recognition. Simulation results demonstrate that the proposed method is effective.