基于关联特征提取的舰船噪声混迭谱分解

Ship Noise Mixing Spectral Decomposition Based on Correlation Feature Extraction

不同舰船目标辐射噪声的噪声平均功率谱具有差异性特征,能在一定程度上反映舰船目标的吨位、航速、类型等。研究舰船辐射噪声信号的特征提取和频谱分解方法,对提高舰船目标的识别能力具有重要意义。传统的舰船辐射噪声关联特征提取采用的是基于定量递归分析的关联维特征提取方法,当在关联特征提取中舰船辐射噪声信号出现奇异吸引子特征时,提取的舰船目标特征产生混迭谱,导致频谱畸变,影响目标识别性能。针对这一问题,提出一种基于预畸变趋化关联特征提取的舰船噪声信号混迭谱分解方法,设计舰船辐射噪声产生与信号源系统模型,进行声传感器布置研究,进行特征提取和混迭谱分解算法改进分析。仿真实验得出,采用该方法进行舰船辐射噪声信号的预畸变趋化关联特征提取,能有效展示舰船辐射噪声信号的内部规律特征,提高对舰船辐射噪声信号的特征提取性能和目标识别精度。

The average power spectrum of different ship radiated noise is different, and it can reflect the ship target in a certain extent as tonnage, speed, etc. Feature extraction and signal spectrum of ship radiated noise decomposition method is researched to improve the ship target recognition ability. Traditional extraction method uses the correlation dimension feature extraction method based on recurrence quantification analysis, when the ship in the related feature extraction radiated noise signals appear strange attractor characteristic, it produces aliasing spectrum, cause the spectrum distortion, and affect the performance of target recognition. Aiming at this problem, a ship noise mixing spectral decomposition algorithm is proposed based on pre distortion trend correlation feature extraction, and the signal source system model is constructed, acoustic sensor layout is designed, feature extraction and mixing spectral decomposition algorithm analysis is obtained. The simulation experiments show that it can reflect the internal characteristic of ship radiated noise signal, it can improve signal of ship radiated noise feature extraction performance and accuracy of target recognition.