To realize the feature extraction of ship-radiated noise in complex ocean environment,the sample entropy is used to extract the features of three types of ship-radiated noise(SRN).Because the sample entropy only analyzes the SRN signal in single scale and it cannot distinguish different types of ships effectively,a new method of SRN feature extraction is presented based on sample entropy and ensemble empirical mode decomposition(EEMD).Firstly,three types of SRN signals are decomposed with EEMD,and the sample entropy of each intrinsic mode function(IMF) is analyzed to select the IMF sample entropy with the highest energy as the feature parameter.Then,by comparing the IMF highest-energy sample entropy of a certain number of the above three types of SRN signals with the sample entropy of SRN signal,it is discovered that the IMF sample entropy with the highest energy is at the same level for similar ship types,but is quite different for different ship types.Test results show that taking the IMF sample entropy with the highest energy as the feature parameter can obtain better separability for ships,compared with SRN sample entropy.%为了实现舰船辐射噪声在复杂海洋环境中的特征提取,采用样本熵对3类舰船辐射噪声(SRN)进行特征提取.针对样本熵只能在单尺度下对原信号进行分析,无法有效区分不同类别舰船,提出了一种将样本熵与集合经验模态分解(EEMD)相结合的舰船辐射噪声特征提取方法.首先对3类不同种SRN信号进行EEMD,对分解后得到的各阶固有模态函数(IMF)的样本熵进行分析,选取更具有区分度的最强IMF样本熵作为特征参数.通过比较一定数量3类SRN的最强IMF样本熵及原SRN样本熵特征参数发现,同类舰船的特征参数基本处于同一水平,不同类型的舰船存在一定差异.试验结果表明,以SRN的最强IMF样本熵作为特征参数相比原SRN样本熵对舰船具有更好的可分性.
展开▼
