大地电磁测深资料遗传算法反演研究及应用  

Research on Genetic Algorithm Inversion for Magnetotelluric Sounding Data and its Application

作　　者：涂齐催[1] 何贤科[1] 姜雨[1] 杜本强 

机构地区：[1]中国海洋石油有限公司上海分公司研究院,上海200030 [2]中石油冀东油田分公司石油勘探开发研究院,河北唐山063004

出　　处：《石油天然气学报》2012年第5期83-86,5,共4页Journal of Oil and Gas Technology

摘　　要：反演是大地电磁测深(MT)的关键环节之一。MT反演是多参数非线性最优化问题,存在多解性;而基于遗传算法的MT反演可提高非线性解的唯一性,它只需要问题的正演公式及给定参数的范围,不需求导数,也不要求有良好的初值,具有抗干扰能力强、拟合度高等优点,而且遗传算法作为一种非线性全局优化方法能在全局范围内搜索最优解。通过建模,在正演的基础上进行了反演研究,得到的结果与模型基本一致。在此基础上,运用遗传算法反演对实测资料进行了处理,效果较好。Inversion was one of the important factors of magnetotelluric sounding(MT).MT inversion was a non-linear problem of multi-parameter optimization and it had several solutions.MT inversion based on genetic algorithm could improve the uniqueness of non-linear solutions.Only the forward formula and range of given parameters were needed,and derivatives and good initial values were not demanded.The method had the advantages of anti-interference and high-fitting.Besides,as a non-linear and optimization method,GA could search for the best solution in a wide range.By building a model,based on forward modeling,MT inversion is studied and the inverted result is in accordance with the model.The MT inversion based on GA is used to process the practical measured data and a satisfied result is obtained.

关 键 词：大地电磁测深 正演公式 遗传算法 模型 

分 类 号：P631.325[天文地球—地质矿产勘探]