一种新的连续自适应格点反褶积方法

在“格点滤波--子波处理的一种新技术”(《石油地球物理勘探》,21(2)1986:111～124)一文中,已经建立了比较完整的格点滤波理论。由此出发,提出一种连续自适应估算格点滤波参数的新方法。实现此方法的关键在于连续自适应地估计偏相关系数K_m。Makhoul和Griffth分别提出的自适应方法都存在以下三个问题:①由于时间方向上采用逐点往前递推的方法,无法使用高速阵列机;②在估计n时刻的Km时,只利用n时刻以前的信息,所以自适应跟踪速度总是慢于信号的时变速度;③不能对信噪比与分辨率作一折衷的选择。本方法由于采用与输入信息有关的和式比来表示K_m,而且宜于以数据流为单位进行运算,所以这种新的自适应估算方法能够采用阵列机运算,实现快速跟踪的连续自适应格点反褶积。在此反褶积方法中引入加白噪的方法,还能有效地控制信噪比和分辨率。理论研究与实际资料处理表明,这种新方法具有简单、易行、快速、效果稳定的特点。

A new continuous adaptive lattice deconvolution

Quite perfect lattice filtering theory has been narrated in the paper Lattice filtering-A new wavelet processing technique published in 《OGP》,21(2)1986:111～124,A new method for continuous adaptive estimation of lattice filtering parameter is advanced on the basis of the lattice filtering theory.The key of this method lies in the continuous adaptive estimation of the partial correlation coefficient KTheadaptive estimation methods advanced respectively by Makhoul and Griffth encounterthree problems: (1) high effective array processor can not be used because of the adoption of point by point forward recursion in time series; (2) when the K_m, at time n is estimated,only the signals before time n are used so that adaptive tracking is always slower than signal variation with time; (3) no compromise between signal to noise ratio and resolution can be taken. This new adaptive estimation method can be performed on array processor to accomplish fast tracking continuous adaptive lattice deconvolution because K_m is given in the form of ratio of summation formulas relating to input informations and data flow may be considered as an unit for operation.If white noise is introduced into the deconvolution,both signal to noise ratio and resolution can be controlled effectively. Both the theory and practice prove this new method to be simple, convenient. highly effective anal stable.