水声信号时频分析方法比较及应用研究  

A Comparison of Time-Frequency Analysis Method and Their Applications to Underwater Acoustic Signal

作　　者：姚玉玲[1] 王宁[1] 石洪华[1] 谭君红[1] 

机构地区：[1]中国海洋大学信息科学与工程学院,山东青岛266100

出　　处：《中国海洋大学学报（自然科学版）》2011年第11期115-119,123,共6页Periodical of Ocean University of China

摘　　要：浅海中的宽带水声信号传播呈现出频散的特点,通过高分辨率的时频分析方法可以刻画频散曲线。通过数值仿真和实验数据处理,对比分析几类常用的时频分析方法在提取宽带声信号频散曲线方面的性能。结果表明:STFT时频局部化精度不够高;在较强频散的情况下,DSTFT时频分辨率较高。WVD时频聚集性最好,但是有严重的交叉项干扰;固定核函数的CWD较好地抑制交叉项,时频分辨率虽优于STFT,但弱化了时频聚集性;AOK时频分布采用自适应高斯核函数,在抑制交叉项的同时,时频聚集性较好,有望较好地用于提取信号频散曲线。An acoustic wave in a shallow water ocean channel is proved dispersive.The dispersive property can be got by high resolution time-frequency analysis method.At present,there are a lot of time-frequency analysis methods.Short time Fourier transform（STFT）,Wigner-Ville distribution（WVD）,Cohen class distribution（CWD）,adaptive optimum kernel time-frequency representation（AOK）,Dispersion-based short-time Fourier transform（DSTFT） are studied in this paper.By using these methods to a simulated acoustic signal and an experiment signal and comparing the properties in time-frequency resolution precision,this paper draws the following conclusion： the time-frequency resolution precision of linear time-frequency representation including STFT is lower.Although the time-frequency resolution precision of WVD is excellent,its cross-terms are rather severe.CWD,the kernel function of which is exponential,can repress the cross-terms to a certain extent and its time-frequency resolution is better than STFT which is much lower than that of WVD.AOK time-frequency representation using adaptive kernel function is a relatively ideal method which can repress the cross-terms and has better time-frequency resolution.

关 键 词：WVD 自适应最优核 DSTFT 

分 类 号：O427.2[理学—声学]