最大熵子波反褶积

最大熵子波反褶积是建立在最大熵反褶积的基础上。最大熵反褶积的优点在于不需要求子波的自相关值和期望输出与子波的互相关值，即不需要解托布里兹方程组求出反滤波因子，而只要求出各i阶偏相关系数ρ_i(i=1，…，M)就可由数据直接递推地求出第M阶向前预测误差和向后预测误差。这种作法可以避免在数据时窗外用补零的办法求相关值而带来的误差。最大熵反褶积是在地震子波为最小相位条件下进行的，其向前预测误差就等于反射系数序列。最大熵子波反褶积与最大熵反褶积的不同点在于，该方法将所得反射系数序列作为初值，用此反射系数序列与地震记录作相关提取单道子波，再采用多道统计平均法求取最终子波。为使其最小相位化，对最终子波要进行整形，然后再作子波反褶积。经模型试验与实际资料处理表明，该方法效果良好。

Maximum entropy wavelet deconvolution

Maximum entropy wavelet deconvolutian is based on maximum entropy deconvolution. The advantage of maximum entropy deconvalution is that we donut solve Toeplitz matrices to obtain inverse filtering operator (namely,without calculating the wavelet autocorrelation values and the values of cross correlation between expected output and wavelet),but only calculate the forward and backward m-order prediction errors by making a direct data recurrence after partial carrelation coefficients ρ_i(i=1,2... m)are calculated. This method can avoid the error which results from the way that carrelxtian values are obtained by null fill-in outside the data window. The maximum entropy deeonvalution is performed under the premise that seismic wavelet is of minimum phase,so the forward prediction error is equivalent to reflection coefficient series. The difference between maximum entropy wavelet deconvolution and maximum entropy deconvolution is that the calculated reflection coefficient series in maximum entropy wavelet deconvolution znethad is used as initial values and correlated with seismic record to obtain single trace wavelets,from which a final wavelet is taken using mufti-trace statistical method,The final wavelet should be shaped into a minimum phase wavelet,which is then used in wavelet deconvolution. Both the modeling and real data processing prove this method satisfactory. entropy deeonvalution is performed under the premise that seismic wavelet is of minimum phase,so the forward prediction error is equivalent to reflection coefficient series。The difference between maximum entropy wavelet deconvolution and maximum entropy deconvolution is that the calculated reflection coefficient series in maximum entropy wavelet deconvolution znethad is used as initial values and correlated with seismic record to obtain single trace wavelets, from which a final wavelet is taken using mufti-trace statistical method, The final wavelet should be shaped into a minimum phase wavelet, which is then used in wavelet deconvolution。Both the modeling and real data processing prove this method satisfactory.