舰船辐射噪声分类识别一直是被动声呐面临的难题，提取舰船辐射噪声的频谱特征来实现分类识别是一种常用的方法。基于舰船辐射噪声频谱特征主要聚于低频段的特点，按照稀疏分解的原理，通过构造完备的非线性频谱字典，提出了一种舰船辐射噪声非线性频谱特征提取方法。对海上实录的多种型号和多种工况的大量噪声样本进行了特征提取，采用最近邻分类器对辐射噪声样本进行了分类识别实验，结果表明，非线性频谱特征的正确分类识别概率高于线性频谱特征的正确分类识别概率。%Classification and recognition of ship radiated noise is always a difficult problem. It is a commonly used method to extract the spectrum features of ship radiated noise. Based on ship radiated noise spectrum characteristics mainly in the low frequency characteristics, in accordance with the principle of sparse decomposition, by constructing completenon-linear spectral dictionary proposed a kind of ship radiation spectrum features of nonlinear noise extraction method. There-cord of sea of various types and various conditions of a lot of noise samples are feature extraction, using the nearestneigh-bor classifier on radiated noise samples were classified recognition experiment. Results show that, nonlinear frequencyspec-trum feature of the probability of correct classification and recognition than linear spectral characteristics of the correctclas-sification probability.
展开▼
