全波震相模糊识别及应用

Fuzzy determination of full wave phases and its application

带时间参量t~*产的Fourier变换只能分辨不同震相的子波频谱,但不能从根本上把不同的子波分开。采用波动理论修正函数及模糊识别方法建立的t~*-φ图,可对全波震相进行识别,其可靠性要比原始记录高。φ表示无量纲隶属函数,取值范围为[0,1]。若φ=1,则认为t~*时刻有子波出现;若φ=0,表示t~*时刻无子波出现;若0【φ【1,则子波出现的确定程度随φ值偏离1的距离减小而增大。波动理论修正函数随振幅修正系数改变,因而可取多种形式,以便达到识别各种震相的目的。选取波动理论修正函数时,要考虑粘弹性介质的频散特性和子波叠加的影响,此法有利于多道分析及对多次波的识别,若和文献[1]提出的方法联合运用,则可进一步突出所识别的子波的频谱特性,改善图像显示效果。

Fourier transform with time parameter t favours identifying the wavelet frequency spectra of different phases, but fails to part different wavelets. t-φ curve that is plotted using both theoretical wave revision function and fuzzy recognition can be applied to the identification of full wave phases, and its reliability is better than that of raw seismic record. φ, ranging from 0 to 1, represents dimensionless membership function, if φ=1, wavelets appear at time t; if φ=0, no wavelet appears at t; and if 0<φ<l, the possibility of wavelet appearance increases with the decrease of difference between φ and 1. Theoretical wave revision function varies with amplitude revision coefficient; therefore, it can take many forms to achieve the identification of all phases. Both the dispersion of viscoelastic medium and the influence of wavelet stack should be considered in determining theoretical wave revision function. This method favours multi-trace analysis and multiple wave identification. Combined with the method described in literature [1], this method can enhance frequency spectrum characteristics of identified wavelets and improve image display.