【摘要】 本文通过引入高阶统计学的独立变量分析算法(ICA),利用双谱估计出的地震子波作为约束条件来实现同态盲地震反褶积。首先在无噪声的条件下,将地震记录由时间域变换到同态域,利用相邻两道地震道作为组合,将地震褶积模型转换为线性混合ICA模型;然后,利用高阶统计学中的双谱估计地震子波,利用估计出的地震子波信息作为约束条件应用于快速独立变量分析算法(FastICA)中,实现地震子波和反射系数在同态域的分离;最后将分离的地震子波和反射系数再反变换到时间域,即可得到反射系数和地震子波。模拟和实际地震数据的数值算例均表明,在不对反射系数作高斯白噪假设及子波最小相位假设的条件下,基于高阶统计约束的同态盲反褶积方法能够取得较好的反褶积效果。更多还原

【Abstract】 A high order statistical method,so called,in-dependent components analysis(ICA)with constraint of seismic wavelet estimated from bispectrum of seismic traces,is introduced in this paper to improve the traditional homomorphic seismic deconvolution method.In the noise-free condition,seismic records are converted from the time domain to the complex cepstrum domain in order to transform the common seismic model to the basic ICA model.By applying FastICA algorithm,reflectivity series and the seismic wavelet can be produced in complex cepstrum domain and converted back to the time domain subsequently.Model and real seismic data numerical examples show that this method can blindly and effectively inverse the wavelet and the reflectivity at the same time without the assumptions of Gauss reflectivity white noise and wavelet minimum phase.The algorithm referred here is an updated version of homomorphic deconvolution methods for seismic signal blind deconvolution and merits more researches.更多还原