摘要
地球物理反问题是一个典型的非线性最优化问题，而常用于评价解估计的目标函数却是一个具有多极值的非线性问题。常规的局部线性化反演方法往往易于陷入局部极大值之中，且严重依赖于初始模型的选取。遗传算法则是一种全局性搜索方法，不易于陷入局部极大值之中，且对初始值依赖性不大，能较好地解决这一问题。本文在遗传算法中引入了多个目标函数的综合评价和灾变过程，对层状弹性介质（水平界面和弯曲界面）的纵、横波速度和密度进行了同时反演，证明遗传算法对于解决地球物理反问题是有效的，且具有较强的抗噪性。
Geophysical inversion problem can be considered as a nonlinear optimization problem. The objective function commonly used in solution evaluation is nonlinear and multiextreme. The inversion methods based on local linearization are usually lost in local maximum values,and they seriously depend on the selection of initial model. The genetic algorithm is a global search one, which is hardly lost in local maximums and has less dependence on initial values.Disaster and comprehensive evaluation of multiple objective functions are introduced into the genetic algorithm. We performed the simultaneous inversions of P,Swave velocities, and density of the layered elastic medium with horizontaI and curved interfaces. It has been proved that the genetic algorithm is very effective on geophysical inversion and quite noise resistant.
出处
《石油地球物理勘探》
EI
CSCD
北大核心
1994年第2期156-165,共10页
Oil Geophysical Prospecting
关键词
遗传算法
弹性介质
反演
地震数据
genetic algorithm, objective function,disaster, layered elastic medium, inversion
