基于SGMD和MIPE的舰船辐射噪声特征提取方法研究

Feature extraction of ship radiated noise based on SGMD and MIPE

摘要:

针对复杂环境下低信噪比水声信号特征提取的难点，本文提出一种基于辛几何模态分解（SGMD）、多尺度改进排列熵（MIPE）舰船辐射噪声特征提取方法。该方法利用SGMD对三类舰船辐射噪声信号进行降噪，有效抑制噪声干扰并保留信号本质特征；提取MIPE特征作为分类的特征参数，并输入概率神经网络（PNN）进行分类识别。实验结果表明，在0dB和5dB低信噪比环境下，所提方法的平均识别率分别达到96.89%和99.56%，相较于未降噪处理前分别提升9.11%和7.79%，验证了该方法在低信噪比下具备良好的特征稳健性与分类性能。

Abstract:

To address the challenge of feature extraction from ship-radiated noise in complex environments with low signal-to-noise ratio (SNR), this paper proposes a method based on Symplectic Geometry Mode Decomposition (SGMD) and Multiscale Improved Permutation Entropy (MIPE). The proposed method utilizes SGMD to denoise the radiated noise signals from three types of ships, effectively suppressing noise interference while retaining the intrinsic signal characteristics. Subsequently, the MIPE features are extracted as classification parameters and fed into a Probabilistic Neural Network (PNN) for classification and recognition. Experimental results demonstrate that under low SNR conditions of 0 dB and 5 dB, the average recognition rates of the proposed method reach 96.89% and 99.56%, respectively. These represent significant improvements of 9.11% and 7.79% compared to the results obtained without the denoising process, thereby validating the excellent feature robustness and classification performance of the proposed method under low SNR conditions.