自适应的最小平方反褶积及在地震勘探中的应用

本文给出了一种自适应的最小平方反褶积方法（ALSD法）,其基本原理是用滤波器的输出来修正愿望输出,从而使结果得到改善。从迭代格式来讲,它是极小熵反褶积方法的推广。本文还从理论上讨论了自适应函数的选取方法。通过人工模拟地震记录数据及真实地震剖面计算的结果,显示出ALSD法是一种计算量省且效果好的方法。对小相位子波,滤波效果与最小平方预测反褶积相当;对混合相位子波,仍具有近于小相位时的效果。

ADAPTIVE LEAST SQUARE DECONVOLUTION AND APPLICATION TO SEISMIC PROSPECTING

This paper presents an adaptive least square predictive deconvolution method (ALSD). Its basic principle is that successively adjusting desired signals according to filtering outputs makes improved results being adapted to the wavelet phase. In iterative algorithm the method generalizes iterative model of minimum entropy deconvolution. The paper gives some criteria for choosing an adaptive function model which is of significance for algorithm. Numerical examples with synthetic data and natural data show an improvement result and less computational amount. For minimum phase wavelet the result can be compared with that of spike deconvolution and for non-minimum phase wavelet the result is close to that of corresponding minimum phase wavelet.