【摘要】 在连续小波基的平稳序列时频分析方法基础上,对非平稳时间序列的时频分析方法深入研究,引入经验模态分解法,利用信号的局部特征——时间尺度,从原信号中提取若干个固有模态函数(IMF)和一个残余量,用多个固有模态函数表征信号的局部特征,用残余分量表征信号缓慢变化的趋势,并形成非平稳信号经验模态分解算法。通过实例应用研究验证了EMD分解后所作的瞬时Hilbert幅值谱,比连续小波变换CWT分解后的谱的分辨率高,并与地震道波形匹配自然,可用于地震非平稳数据的时频分析。更多还原

【Abstract】 In this paper,the frequency analysis method for non- stationary time series achieved further study based on stationary sequence frequency of continuous wavelet analysis method. We introduced empirical mode decomposition method which uses local characteristic time scale signal to extract a number of intrinsic mode function( IMF)from original signal and a residual amount,and based on the local features of several intrinsic mode function signals,the non- stationary signals empirical mode decomposition algorithm was set up according to the trends of slow component of the residual signal. Application practice verified that after the EMD Hilbert instantaneous amplitude spectrum than by continuous wavelet transform CWT resolution spectral decomposition,and naturally matched the seismic trace of waveform. These confirmed that the proposed method can be used for non- stationary frequency seismic analysis.更多还原