STFT与WVD在地震波时频分析中的应用  

Application of short-time Fourier transform and Wigner-Ville distribution in time-frequency analysis of seismic wave

作　　者：刘希康[1,2] 丁志峰[2] 李媛[1] 崔仁胜[3] 

机构地区：[1]中国地震局第一监测中心,中国天津300180 [2]中国地震局地球物理研究所,中国北京100081 [3]中国地震局地震预测研究所,中国北京100036

出　　处：《地震地磁观测与研究》2016年第1期15-21,共7页Seismological and Geomagnetic Observation and Research

基　　金：中国地震局"三结合"(No.163303);中国地震局第一监测中心主任基金项目(FMC2015005);科技部公益性行业科研专项"中国地震科学台阵探测--南北地震带北段"(No.201308011)

摘　　要：地震波是一种非平稳信号,传统的傅里叶变换已不适用,时频分析方法能在时间—频率域给出非平稳信号不同时刻的频率分布特征。Wigner-Ville分布(WVD)对非平稳信号具有较好的时频聚集性,但因其为双线性变换,会产生严重的交叉干扰项。短时傅里叶变换(STFT)是一种线性变换,不存在交叉项干扰,但时频聚集性较差。考虑到二者优点,在对地震波进行时频分析时,采用短时傅里叶变换和Wigner-Ville分布的联合算法,既有效抑制了Wigner-Ville分布交叉项干扰,又保持了较高的时频聚集性。通过3种方法,对模拟信号和实际数据处理结果进行对比分析,发现联合算法对获取地震波信号的时频特征和纵波到时检测,具有更好效果。Seismic signal is a kind of non-stable signal, which can＇t be analyzed suitably by classical Fourier analysis. Time-frequency analysis can analyze non-stationary signal in time-frequency domain and show out the distribution characteristics. WVD has perfect time-frequency concentration, but because it is a bilinear transformation, the existence of cross-terms is inevitable. STFT is linear transformation, has no cross-terms, but affected by the window size,it has bad time-frequency concentration. Based on the advantage of these two methods, we propose to obtain the time-frequency distribution by the combined method. This method can suppress the cross-term interference in WVD effectively, while keeping the excellent timefrequency concentration. By calculating simulation signal and actual data, the comparison result of STFT, WVD and the combined method indicates that the combination method is most suitable for computing the time-frequency distribution characteristics of earthquake waveform signal and more effective to detect the arrive time of P-waves.

关 键 词：短时傅里叶变换 WIGNER-VILLE分布 时频分析 地震波谱 

分 类 号：P315.31[天文地球—地震学]