舰船辐射噪声降噪是水声信号处理的基础，为了获得更好的降噪效果，将基于互补集合经验模态分解（CEEMD），提出一种结合排列熵（PE）、小波软阈值（WST）降噪和奇异谱分析（SSA）的联合降噪方法。该方法首先通过互补集合经验模态分解将含噪信号分解为一系列本征模态函数，然后用排列熵对有效模态分量和含噪模态分量进行区分，对含噪模态分量进行小波阈值去噪后和有效模态分量进行重构，最后对重构信号利用奇异值分析方法进一步提取有效成分后得到降噪后的信号。将所提方法用于仿真数据、混沌信号和实测舰船辐射噪声进行实验，实验结果验证了所提出方法的可行性和有效性。
Ship radiation noise reduction is fundamental to underwater acoustic signal processing. To achieve better noise reduction effects, this paper proposes a joint denoising method based on Complementary Ensemble Empirical Mode Decomposition (CEEMD), combined with Permutation Entropy (PE), Wavelet Soft Thresholding (WST), and Singular Spectrum Analysis (SSA). The method first decomposes the noisy signal into a series of Intrinsic Mode Functions (IMFs) through CEEMD. Then, it distinguishes between effective modal components and noisy modal components using permutation entropy. After applying wavelet threshold denoising to the noisy modal components, they are reconstructed with the effective modal components. Finally, the reconstructed signal is further processed using singular value analysis to extract effective components, resulting in the denoised signal. The proposed method was tested on simulated data, chaotic signals, and actual ship radiation noise. The experimental results verify the feasibility and effectiveness of the proposed method.
2026,48(2): 114-121 收稿日期：2025-4-15
DOI：10.3404/j.issn.1672-7649.2026.02.019
分类号：U661.4；TN911.7
基金项目：广西研究生教育创新计划项目（YCSW2024344）
作者简介：庄泽文(1999-),男,硕士研究生,研究方向为水声信号处理
参考文献：
[1] 倪俊帅, 赵梅, 胡长青. 基于DNN和改进K-means的船舶辐射噪声开集识别方法[J]. 声学技术, 2022, 41(3): 382-387.
NI J S, ZHAO M, HU C Q. Open-set recognition method for ship radiated noise based on DNN and improved K-means[J]. Technical Acoustics, 2022, 41(3): 382-387.
[2] 李紫鹏, 纪永强, 郭兵勇, 等. 基于特征模式分解的水声目标特征提取方法[J]. 哈尔滨工程大学学报, 2023, 44(9): 1542-1548.
LI Z P, JI Y Q, GUO B Y, et al. Feature extraction method for underwater acoustic targets based on feature mode decomposition[J]. Journal of Harbin Engineering University, 2023, 44(9): 1542-1548.
[3] 郭帆, 王鹏. 基于集合经验模态分解的MEMS矢量水听器联合去噪算法[J]. 测试技术学报, 2022, 36(2): 117-121+177.
GUO F, WANG P. Joint denoising algorithm for MEMS vector hydrophones based on ensemble empirical mode decomposition[J]. Journal of Test and Measurement Technology, 2022, 36(2): 117-121+177.
[4] ZHANG Y, YANG Z, DU X, et al. A new method for denoising underwater acoustic signals based on EEMD, correlation coefficient, permutation entropy, and wavelet threshold denoising[J]. Journal of Marine Science and Application, 2024, 23(1): 222-237.
[5] 孙晓娟, 李建卓. 基于CEEMD-PE的脑电信号降噪方法研究[J]. 电子设计工程, 2021, 29(1): 20-23+29.
SUN X J, LI J Z. Denoising method for EEG signals based on CEEMD and permutation entropy[J]. Electronic Design Engineering, 2021, 29(1): 20-23+29.
[6] 刘倩, 李梅. 基于CEEMD-MI的目标回波参数估计[J]. 舰船科学技术, 2023, 45(22): 155-159.
LIU Q, LI M. Target echo parameter estimation based on CEEMD and mutual information[J]. Ship Science and Technology, 2023, 45(22): 155-159.
[7] 孙润发, 汤占军. CEEMD与AO-SVM结合的风机轴承故障诊断[J]. 机械科学与技术, 2024, 43(6): 962-966.
SUN R F, TANG Z J. Wind turbine bearing fault diagnosis based on CEEMD and AO-SVM[J]. Mechanical Science and Technology for Aerospace Engineering, 2024, 43(6): 962-966.
[8] 韩家鹏, 丁蕾蕾, 韩崇. 基于奇异谱分析的Transformer神经网络光伏预测算法[J]. 软件导刊, 2025, 24(3): 8-15.
HAN J P, DING L L, HAN C. Photovoltaic prediction algorithm based on singular spectrum analysis and Transformer neural network[J]. Software Guide, 2025, 24(3): 8-15.
