舰船辐射噪声的特征提取是水下识别的依据，传统的特征提取可识别性较弱，水下识别较为困难。本文提出一种基于变分模态分解（variational mode decomposition, VMD）与改进多尺度加权排列熵（improved multisacle weighted permutation entropy, IMWPE）相结合的方法进行特征提取，将原始信号通过VMD分解成多个固有模态函数（intrinsic mode function, IMF），选取能够充分体现目标复杂度特征的IMF作为研究对象，然后通过IMWPE方法采用平移均值法解决多尺度加权排列熵（multisacle weighted permutation entropy, MWPE）的单一粗粒化问题。实验数据表明，将本文算法与对比算法提取的特征参数经过粒子群优化的支持向量机（particle swarm optimization support vector machine, PSO-SVM）进行分类识别，IMWPE算法识别率最高，具有良好的稳定性和优越性。
The feature extraction of ship radiated noise is the basis of underwater recognition. The traditional feature extraction is less recognizable, and underwater recognition is more difficult. This paper proposes a feature extraction method based on variational modal decomposition and improved multisacle weighted permutation entropy. The original signal is decomposed into multiple intrinsic modal functions through VMD, and the IMF that can fully reflect the characteristics of the target complexity is selected as the research object. Then the IMWPE method uses the translation mean method to solve the single coarse-grained problem of multi-scale weighted permutation entropy. The experimental data show that the feature parameters extracted by the algorithm in this paper and the comparison algorithm are classified and recognized by the particle swarm optimization support vector machine. The IMWPE algorithm has the highest recognition rate and has good stability and superiority.
2023,45(4): 121-127 收稿日期：2022-03-08
DOI：10.3404/j.issn.1672-7649.2023.04.024
分类号：U666
基金项目：国家自然科学基金资助项目（61901079）
作者简介：丁元明(1967-),男,博士,教授,研究方向为水下信号处理
