A New Denoising Method for Ship-Radiated Noise Based on Complete Ensemble Empirical Mode Decomposition

Noise reduction lays a foundation for feature extraction and identification of ship-radiated noise. To improve the signal-to-noise ratio (SNR) of ship-radiated noise, a new denoising scheme is proposed in this paper, based on complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and improved permutation entropy (IPE). The proposed approach initiates with the decomposition of ship-generated acoustic signals into multiple intrinsic mode functions (IMFs) through the CEEMDAN technique. Subsequently, these IMF components are categorized into three distinct groups -pure noise components, noise-biased components, and signal-dominant components - based on their respective IPE metrics. The processing strategy involves discarding the pure noise IMFs while applying least mean square adaptive filtering (LMSAF) to refine the noise-biased components. The final denoised output is reconstructed by integrating the processed noise-dominant IMFs with the unaltered signal-dominant IMFs. Experimental validation through both simulated and real-world acoustic datasets demonstrates the superior performance of this approach. Comparative analysis reveals that the proposed technique outperforms conventional methods, providing a more reliable preprocessing solution for underwater acoustic signal analysis and subsequent interpretation tasks.