地球物理学进展 ›› 2014, Vol. 29 ›› Issue (5): 2060-2065.doi: 10.6038/pg20140510

• 固体地球物理学 • 上一篇    下一篇

改进的自适应模拟退火密度界面反演方法

秦静欣1, 郝天珧2, 郭子祺1, 乔彦超1, 刘建英1   

  1. 1. 中国科学院遥感与数字地球研究所, 北京 100101;
    2. 中国科学院地质与地球物理研究所, 北京 100029
  • 收稿日期:2013-12-12 修回日期:2014-07-17 出版日期:2014-10-20 发布日期:2014-10-20
  • 通讯作者: 郭子祺,男,1963年生,主要研究方向:无人机航磁探测与解释,环境监测等.(E-mail:guozq@radi.ac.cn) E-mail:guozq@radi.ac.cn
  • 作者简介:秦静欣,女,1982年生,博士。主要研究方向:综合地球物理研究.(E-mail:qinjx@radi.ac.cn)
  • 基金资助:

    中国地质调查局资助项目(GZH200900504)、国土资源部深部探测技术与实验研究专项(90814014,201011080-02)、国家自然基金项目(4121005,41074058,90814011)、国家863项目课题(2009AA09340)和油气专项(2011ZX05008-006-30)联合资助.

The density interface inversion method of improved adaptive simulated annealing

QIN Jing-xin1, HAO Tian-yao2, GUO Zi-qi1, QIAO Yan-chao1, LIU Jian-ying1   

  1. 1. Institute of Remote Sensing and Digital Earth Chinese Academy of Sciences, Beijing 100101, China;
    2. Institute of Geology and Geophysics Chinese Academy of Sciences, Beijing 100029, China
  • Received:2013-12-12 Revised:2014-07-17 Online:2014-10-20 Published:2014-10-20

摘要:

随着计算能力的提高和非线性反演算法的改进,应用非线性反演方法来解决实际物理问题逐渐增多.在众多的非线性密度界面反演方法中,自适应模拟退火反演方法便是其中之一.它易添加约束,不需要计算目标函数的偏导数和大矩阵方程,从而可以很容易找到一个全局最优解.然而,这种方法不适用于一些缺乏相应的数据或无约束信息的区域,因此本文提出了一种改进的自适应模拟退火密度界面反演方法.该方法通过引入变剩余密度建模来不断约束初始模型,在反演过程中通过迭代反演各矩形单元体的密度参数以及深度模型,直到满足收敛条件,最终得到理想的深度界面形态以及剩余密度分布.

Abstract:

With the increase in computing power and nonlinear inversion algorithm improvements, nonlinear inversion methods to solve practical physical problems gradually increased. Among the non-linear density interface inversion method, adaptive simulated annealing inversion method is one of them. It is easy to add constraints. And it doesn't need to calculate partial derivative of the objective and large matrix equation. So you can easily find a global optimal solution. However, this method is not suitable for some lack of the corresponding data or unconstrained information area. Therefore, we propose an inversion of density interface simulated annealing improved adaptive in this paper. The method keeps the remaining constraints initial model by introducing a variable density model. In the inversion process, we adopted an iterative inversion density and the depth of the model parameters for each rectangular element, until it meets the convergence condition. Finally we'll get the desired depth of the interface morphology and density distribution of the remaining.

中图分类号: