摘要
鉴于全波形反演的非线性和多解性,其反演容易陷入局部极小值;在低频缺失情况下,常规波形反演方法对低频信息的恢复能力较弱且波形反演产生海量计算。采用多尺度方法可减弱波形反演对低频信息的依赖性,使反演解不易陷入局部极小值;采用多重网格法可节约计算成本。在多重网格法多尺度波形反演的基础上,对梯度优化方法进行研究,提高反演精度。模型处理结果证明多重网格法多尺度波形反演能够大幅减小计算量,反演解不易陷入局部极小值,梯度优化方法可有效提高反演精度。
In view of the non-linearity and multi-solution of the full waveform inversion,the inversion solution is easy to fall into local minimum.In the case of low frequency loss,conventional waveform inversions are weak to recover low frequency information,and the waveform inversion leads massive calculations.On one hand the multi-scales can reduce the dependency of the waveform inversion on the lowfrequency information,so that the inversion solution is not easy to fall into the local minimum.On the other hand the multi-grids can reduce the cost calculation.We adopt the gradient optimization in the multi-grid and multi-scale waveform inversion to improve the inversion accuracy.Model tests show that the multi-grid and multi-scale waveform inversion can not only reduce the computational complexity,but also avoid local minimum.The proposed gradient optimization can effectively improve the inversion accuracy.
出处
《石油地球物理勘探》
EI
CSCD
北大核心
2017年第6期1208-1217,共10页
Oil Geophysical Prospecting
基金
国家自然科学基金项目(41374122
41504100)
阿美亚洲&创行"能源
环境及气候可持续发展"项目
山东省自然基金(ZR2017MD014)
中国石油大学(华东)主要创新科研计划项目(17CX02052)资助
关键词
能量补偿
波印廷矢量
梯度分解
多重网格法
多尺度
energy compensation
Poynting vector
gradient decomposition
multi-grids
multi-scales
