Algorithm of complex wavelet cluster for extracting PD signal

In order to accurately detect partial discharge of power equipment, white noise must be suppressed. In this paper, according to the characteristic of partial discharge ( PD ) signal and the complex wavelet coefficients of white noise and that the nature of wavelet transform is to calculate the inner product of the signal and wavelet, the algorithm in which an optimum complex wavelet is selected from a cluster of complex wavelets is put forward. Based on the threshold de-nosing algorithm, an algorithm of complex wavelet cluster is given. In this algorithm, the decompose level is decided by the theory of modulus minimum. Then the white noise in polluted PD signal is suppressed by this algorithm, and the de-noising effect is compared with the effect by the algorithm of single complex wavelet threshold. At the same time, the de-noising effect of the emulation PD signal is evaluated by Signal to Noise Ratio (SNR) and Normalized Correlation Coefficient (NCC). If complex wavelet cluster method is applied, its SNR is more than 10 dB, and its NCC is more than 0.9. However, if single complex wavelet threshold method is applied, its SNR is lower than the former, and its NCC is just more than 0.73. The results indicate that the de-noising effect of complex wavelet cluster is better than that of single complex wavelet and it can suppress white noise effectively with small distortion and high SNR.