Denoise for propeller acoustic signals based on the improved wavelet thresholding algorithm of CEEMDAN

AbstractThe primary objective of this study is to address the issue of noise in underwater propeller signals. The motivation for this research arises from the challenge of distinguishing signal and noise components in underwater propeller signals, particularly in the presence of complex hydrodynamic interactions. So it introduces an improved wavelet threshold algorithm based on CEEMDAN for noise reduction in underwater propeller signals. The algorithm utilizes the energy concentration property of wavelet transform to distinguish signal and noise components. And by setting the noise coefficients based on the Balanced Wavelet Thresholding (BWT) algorithm, effective denoising is achieved. Compared to the traditional wavelet threshold method, the study evaluated three approaches, EMD-CWT, EMD- BWT, and CEEMDAN-BWT for denoising numerical simulation and experimental signals of propeller cavitation wake acoustic pressure generated using CFD software. The findings show that the CEEMDAN-BWT outperforms others in noise reduction. The research results indicate that the CEEMDAN-BWT algorithm performs superiorly in noise reduction. The Signal-to-Noise Ratio (SNR) has been improved by 2.4326 decibels, and the Root Mean Square Error (RMSE) has increased by 0.0893 decibels. The algorithm effectively preserves signal components, enhancing SNR without introducing excessive distortion, demonstrating its potential in significantly reducing propeller signal noise levels and improving signal quality.