Wavelet based harmonic parameter estimation in non-Gaussian impulsive noise environments

This paper focuses on the harmonic estimation procedure based on continuous wavelet transform (CWT) when the stationary signal to be analyzed is corrupted by the non-Gaussian impulsive noise. CWT of the harmonic signal and the impulsive noise are compared, and it is proved that the parameters of the complex Morlet wavelet have significant impacts on the CWT coefficients. We calculate the total duration of the regions in wavelet ridge that is contaminated by the impulsive noise and the edge effect, and then the optimum wavelet parameters are determined when the duration achieves its minimum value. After the regions are removed, the amplitude, frequency and phase of each harmonic component can be estimated accurately. The effectiveness of the proposed method is confirmed through numerical simulation.