摘要
为实现强海洋背景噪声中的微弱船舶轴频电场信号检测,提出了一种结合经验模态分解(Empirical Mode Decomposition,EMD)和窄带子区间功率谱熵的线谱提取新算法。首先,利用EMD方法从含噪信号中分解出一组有效固有模态函数(Intrinsic Mode Functions,IMFs),对各有效IMF的功率谱进行子区间划分;其次,定义并计算各子区间的能量峰值熵比(Energy Peak Entropy Ratio,EPER)特征;最后,通过对轴频信号和环境噪声物理特征差异的分析,结合K-均值聚类方法进行特征量的筛选,实现线谱提取。海上实测数据的处理结果表明,相比于直接的功率谱分析,算法的线谱可提取下限降低了6.7 d B。
In order to implement detection of weak ship shaft-rate electric field signal buried in strong marineback-ground noise, a new line spectrum extraction algorithm combining Empirical Mode Decomposition (EMD) and narrowband sub-interval power spectra entropy is proposed. Firstly, a set of effective Intrinsic Mode Functions (IMFs) were separated from noise-polluted signal by means of EMD method, of which the power spectra was then divided into subinterval sections. Furthermore, a new feature named Energy Peak Entropy Ratio (EPER) of each section was defined and computed. Finally, taking advantage of analyzing differences of physical properties between shaft-rate signal and ambient noise, together with K-means clustering method, line spectrum was extracted. Processing results of sea measured data indicate that, comparing to direct power spectra analysis, the line spectrum detection low bound is reduced by 6.7 dB.
出处
《舰船科学技术》
北大核心
2017年第9期159-163,共5页
Ship Science and Technology
基金
国家自然科学基金资助项目(51109215
51509252)
关键词
轴频电场
经验模态分解(EMD)
K-均值聚类
能量熵
线谱提取
shaft-rate electric field
empirical mode decomposition (EMD)
K-means clustering
energy entropy
line spectrum extraction