基于神经网络的面波迭代反演应用研究  

STUDY ON THE APPLICATION OF ITERATIVE INVERSION OF SURFACE WAVE BASED ON ARTIFICIAL NEURAL NETWORK

作　　者：贺懿[1,2] 张进[1] 刘怀山[1] 

机构地区：[1]中国海洋大学海洋地球科学学院,山东青岛266100 [2]中海石油有限公司湛江分公司研究院,广东湛江524057

出　　处：《西南石油大学学报（自然科学版）》2010年第1期40-44,共5页Journal of Southwest Petroleum University(Science & Technology Edition)

基　　金：国家"863"项目(2006AA09Z339;2006AA06A108);山东省自然科学基金项目(Y2006E09)

摘　　要：根据滩浅海近地表结构特征,尝试利用地震记录中的面波进行近地表结构研究,以便了解滩浅海地区的近地表地层介质结构变化,为深层油气勘探提供准确的低降速带资料。针对面波频散反演已有方法存在的不足,引入一种基于BP神经网络的迭代反演方法对面波的频散曲线进行拟合迭代,用于反演预测滩浅海低降速带地层参数。由于神经网络具有很强的自学习、自适应、自组织和容错能力,它的反演预测能力非常强大,能够较精确地预测出所要求解的目标数据,同时结合传统迭代反演方法的优点,增强了该方法的反演预测能力。通过对滩浅海近地表结构模型试算,获得好的效果,同时进一步对实际记录进行了计算,也取得了比较满意的结果。Because traditional methods applied in investigating surface-structure are restricted on paralic zone,according to the feature of surface-structure on paralic zone,the try is done to use the surface wave on seismic record to research the surface-structure and to offer deep exploration activity of weathering zone.In view of shortages in the methods applied in dispersion curve inversion of surface wave,an iterative inversion method based on BP(Back-propagation) artificial neural network is introduced to surface wave and it is used to predict the parameters of weathering zone on paralic zone.Combined with very strong self-learning,self-adapting,self-organizing and fault-tolerant ability of neural network,the prediction power of conventional iterative inversion method is enhanced effectively.By testing the model of paralic surface-structure,the good effect can be obtained.Moreover,applied to real data,the method still gives out satisfactory result.

关 键 词：滩浅海近地表结构 面波 频散曲线 BP神经网络 迭代反演 地震勘探 

分 类 号：TE132.1[石油与天然气工程—油气勘探] P631.4[天文地球—地质矿产勘探]