重磁数据的重加权正则化共轭梯度法约束反演  

Re-weighted regularized conjugate gradient constrained inversion of gravity-magnetic data

作　　者：吴昊天 李军[1,2] 王堃鹏[1,2] 李霞[1] WU Haotian;LI Jun;WANG Kunpeng;LI Xia(Chengdu University of Technology College of Geophysics,Chengdu 610059,China;Key Laboratory of Earth Exploration and Information Technology of Ministry of Education,Chengdu 610059,China)

机构地区：[1]成都理工大学地球物理学院,成都610059 [2]成都理工大学教育部地球探测与信息技术重点实验室,成都610059

出　　处：《物探化探计算技术》2021年第2期232-241,共10页Computing Techniques For Geophysical and Geochemical Exploration

基　　金：国家重点研发计划(2017YFC0601504)。

摘　　要：重磁异常的物性线性反演多是欠定问题求解,求解不仅具有明显的不稳定性,还会导致其垂向分辨率大大降低,反演的“趋肤效应”明显。针对重力数据反演中的这一问题,引入重加权正则化共轭梯度法的约束反演,即在目标函数中加入正则化方程加以约束,以深度加权函数对核矩阵进行补偿,在反演过程中实现了正则化因子的自适应选择。引入了PRP公式,改善了反演的计算效率和计算精度。引用一个相对复杂的Y型岩脉的重磁加噪数据,对比分析了FR公式和PRP公式下的重力反演和磁异常反演的结果,并对马角坝3号地质剖面磁异常数据进行了磁反演,论证了算法的有效性和可靠性。The linear inversion of physical properties of gravity and magnetic anomalies is mostly the solution of underdetermined problems.The solution not only has obvious instability,but also causes its vertical resolution to be greatly reduced,that is,the"skin effect"of the inversion is obvious.In order to solve the problem of gravity data inversion,the constraint inversion of reweighted regularization conjugate gradient method is introduced,that is,a regularization term is added to the objective function to constrain,and the kernel matrix is compensated by the depth weighting function,and the adaptive selection of the regularization factor is realized in the inversion process.And the PRP formula is introduced to improve the calculation efficiency and accuracy of the inversion.The gravity-magnetic noise data of a relatively complex y-shaped dyke are quoted,and the results of gravity inversion and magnetic inversion under the FR formula and PRP formula are compared and analyzed,and the magnetic anomaly data of Majiaoba No.3 geological profile was inverted to demonstrate the validity and reliability of the algorithm.

关 键 词：重磁数据 深度加权函数 正则化 共轭梯度法 约束反演 PRP公式 

分 类 号：P631.2[天文地球—地质矿产勘探]