CSAMT一维全区Levenberg-Marquardt反演方法

本文将Levenberg-Marquardt这一全新反演方法引入CSAMT勘探的全区数据反演,提高了CSAMT勘探的深度,避免了测深数据的浪费。Levenberg-Marquardt反演的核心是引入变阻尼的思想,在反演的过程中根据向量各个分量的收敛速度的不同,给定不同的阻尼因子,实现反演的自适应化,这样就大大增加了整个反演的收敛速度和反演精度。反演中采用改进的Bostick变换作为初始模型,减少了反演的迭代次数。

Levenberg-Marquardt Inversion method for CSAMT one-dimensional region data

This articl will introduce a new inversion method which called Levenberg-Marquardt to inverse the CSAMT region data. It increases the exploration depth and avoids the waste of data. Levenberg-Marquardt inversion is the core idea of the introduction of variable damping in the inversion process, and in accordance with the various components of the vector convergence speed of different given different damping factor to achieve inversion adaptive technology, thus greatly increasing the throughout the inversion convergence speed and retrieval accuracy. Inversion using improved Bostick transform as the initial model, reducing the number of iterations of the inversion.