Wavelet Multi-Resolution Analysis Used for Partial Discharge Pattern Recognition

Partial discharge pattern recognition is considered as an effective tool for insulation fault diagnosis on high voltage electric apparatus. Based on the multi-resolution theory of wavelet, a new image recognition method is presented in this paper, which is used in gray intensity image recognition formed by partial discharge phase resolved distribution pattern. Firstly, the outstanding recognition image of partial discharge and the reference one are decomposed by wavelet, and then pattern similarity-degree and adjacency-degree are calculated from the correlation coefficients of low-frequency sub-image at the different scale. The discharge samples extracted from discharge model experiment are processed to pattern recognition by maximum pattern adjacency-degree. According to the recognition rate of multi-resolution image decomposition with four typical wavelet function, the influence of wavelet decomposition scale, wavelet orthogonality and its smoothing continuity to recognizing rate is analyzed. The results show that a satisfied recognition effect has reached through the new method by choosing orthogonal wavelet and suitable decomposition scale