Research on Fault Data Wavelet Threshold Denoising Method Based on CEEMDAN

In order to carry out fault data denoising effectively, this paper proposes a wavelet threshold denoising method based on Complete Ensemble Empirical Mode Decomposition with the Adaptive Noise (CEEMDAN). This method uses CEEMDAN decomposition to obtain a series of frequency from high to low IMF component and the trend term of the fault data; Using permutation entropy value to determine which containing more noise component; using wavelet threshold denoising method to denoise the IMF component of containing more noise, to retain the effective information in the high frequency IMF component; Finally, reconstruction the signal by adding the high frequency IMF component after denoising, low frequency component and the trend term to obtain the denoised data. In this paper, through simulation and measured data to verify this method. The results shows that the proposed method can suppress the noise interference, retain the useful fault signal, extract fault signal with high accuracy effectively.