基于自适应高斯核函数时频分布的水声信号处理研究

Research of Underwater Acoustic Signal Processing Based on Adaptive Gaussian Kernel Time Frequency Distribution

时频分析被广泛应用于水下目标宽带回波信号以及短时瞬态信号的处理。在水声信号的特征检测和分类方面,时频分析的时变谱提供了区分不同类别目标信号的特征信息。提出一种基于自适应核函数时频分布的水声信号处理方法。与固定核函数时频分布相比,自适应高斯核函数时频分布由于它的核函数随信号而自适应改变,因此对交叉项有很好的抑制效果。计算结果表明,对于多分量线性调频水声信号,采用高斯核函数自适应时频分析不仅对交叉项具有很好的抑制作用,而且对信号的自分量具有较好的聚集作用。

There is currently an interest in application of time-frequency analysis to broad band underwater target echo signals processing and short duration transient acoustic events detection which exhibit non-stationary natures. Particularly in the field of feature detection and classification of underwater acoustic signals, the time varying spectrum of time frequency analysis gives more useful information to discriminate different types of target signals. An underwater acoustic signal processing method was put forward based on the adaptive Gaussian kernel time frequency analysis. Compared with fixed kernel distribution, adaptive Gaussian kernel can suppress cross-terms which exist inherently infixed kernel distribution. The simulation results show that the new method not only suppresses the cross-terms well but also concentrates the auto-terms for underwater acoustic signal.