基于交替分裂Bregman迭代算法的鲁棒多道预测反褶积方法

发布时间：2018-06-05 20:46

本文选题：预测反褶积 + L_范数最小化约束　；参考：《石油地球物理勘探》2017年04期

【摘要】：将一次波的L_1范数最小化约束引入多道预测反褶积,提出基于交替分裂Bregman迭代算法的鲁棒多道预测反褶积方法。所提方法利用距离算子求解L_1范数最小化优化问题,在整个迭代过程中只需计算一次矩阵求逆,计算复杂度较低。首先介绍了多道预测反褶积方法的数学模型,然后给出鲁棒多道预测反褶积的优化问题,并阐述了交替分裂Bregman迭代算法求解优化问题的步骤。相对于基于迭代重加权最小二乘法的鲁棒多道预测反褶积方法,文中方法在保持多次波压制效果的同时,能进一步提高计算效率;相对于基于最小二乘法的多道预测反褶积方法和基于交替分裂Bregman迭代算法的鲁棒单道预测反褶积方法,文中方法能有效地均衡一次波的保护和多次波的压制。另外,所提方法利用了多道预测反褶积方法的优势,比单道预测反褶积方法能更好地适应海底的起伏变化。模型数据和实际数据测试结果表明:当水层多次波具有周期性时,文中方法能在保护一次波的同时,有效地压制水层多次波,并具有较高的计算效率;当水层多次波的周期性假设得不到很好的满足时,很难对多次波的压制效果进行直观判断。
[Abstract]:A robust multichannel predictive deconvolution method based on alternating splitting Bregman iterative algorithm is proposed by introducing L-1 norm minimization constraints of primary waves into multichannel predictive deconvolution. The proposed method uses the distance operator to solve the L _ S _ 1 norm minimization optimization problem. In the whole iteration process, it is only necessary to calculate the inverse of the first order matrix, and the computational complexity is relatively low. In this paper, the mathematical model of multichannel predictive deconvolution is introduced, then the optimization problem of robust multichannel predictive deconvolution is presented, and the steps of alternating splitting Bregman iterative algorithm to solve the optimization problem are described. Compared with the robust multichannel predictive deconvolution method based on iterative reweighted least square method, the proposed method can further improve the computational efficiency while maintaining the multiple suppression effect. Compared with the multichannel predictive deconvolution method based on the least square method and the robust single-channel predictive deconvolution method based on alternating splitting Bregman iterative algorithm, the proposed method can effectively equalize the protection of primary wave and suppression of multiple waves. In addition, the proposed method takes advantage of the multi-channel predictive deconvolution method, which is better than the single-channel predictive deconvolution method to adapt to the fluctuation of seafloor. The test results of model data and actual data show that when the water layer multiples are periodic, the proposed method can effectively suppress the water layer multiple waves while protecting the first wave, and has a high computational efficiency. When the periodic hypothesis of water layer multiples is not well satisfied, it is difficult to judge directly the suppression effect of multiple waves.
【作者单位】：中国石油大学(华东)地球科学与技术学院地球物理系;海洋国家实验室海洋矿产资源评价与探测技术功能实验室;
【基金】：中国博士后科学基金面上项目(2015M582163) 山东省自然科学基金项目(ZR2016DB09) 青岛市博士后人员应用研究项目(2015251) 中国石化地球物理重点实验室开放研究基金项目(33550006-15-FW2099-0023) 国家自然科学基金项目(41374122) 国家科技重大专项(2016ZX05026-002-002) 国家重点研发计划项目(2016YFC060110501)联合资助
【分类号】：P631.44

【参考文献】