摘要
自适应正则化反演对初始模型要求较低,直接对瞬变电磁响应数据进行反演。正则化因子通过计算每次迭代的数据目标函数和模型目标函数自适应的得到,从而快速获得地下的地电结构。本文采用自适应正则化反演算法对大定源回线瞬变电磁一维层状模型进行反演,使用均匀半空间作为初始模型,首次采用模型对深度的二阶导数极小的模型约束,通过典型理论模型的反演计算,证明了TEM自适应正则化一维反演算法拟合效果好、精度高,由于不用反复搜索正则化因子,并且收敛速度快,体现了良好的稳定性和可靠性。
Adaptive regularization inversion has less requirement of the initial model,and directly inverts transient electromagnetic response data. In the 1D inversion algorithm,regularization factors are adaptively obtained by calculating the data object function and model object function in each iteration,so that the geoelectric structure in the subsurface can be quickly determined. This paper adopts the adaptive regularization inversion to compute an one-dimensional layered model of transient electromagnetic sounding with a large fixed source loop,using a homogeneous half-space model as the initial model. The model object function is employed to the depth of the second derivative of tiny model constraints for the first time. By calculating the typical theory model,the result of inversion fits very well and has high precision. Furthermore,avoiding repeated search for regularization factors,this method has fast convergence and shows good stability and reliability.
出处
《地质与勘探》
CAS
CSCD
北大核心
2015年第2期360-365,共6页
Geology and Exploration
基金
国家自然科学基金重点基金(U1262206)资助
关键词
瞬变电磁
大定源
自适应正则化反演
最光滑模型约束
TEM,large fixed source,adaptive regularized inversion,smoothest constraint model
