基于主成分分析的自适应宽频带瞬时谱数据融合  

Broad frequency band instantaneous spectral auto-adaptive data fusion based on principal components analysis

作　　者：陈学华[1,2] 贺振华[1,2] 钟文丽[3] 杨俊 

机构地区：[1]成都理工大学油气藏地质及开发工程国家重点实验室,四川成都610059 [2]成都理工大学地球物理学院,四川成都610059 [3]成都理工大学地球科学学院,四川成都610059 [4]中海石油研究总院,北京100027

出　　处：《石油地球物理勘探》2011年第4期576-580,667+495,共5页Oil Geophysical Prospecting

基　　金：国家自然科学基金(41004054)资助项目;高等学校博士学科点专项科研基金(20105122120002)资助项目;四川省教育厅自然科学重点项目(092A011)资助

摘　　要：为了从瞬时谱分解的大量共频率属性数据中提取和突出与储层有关的主要信息,本文提出一种基于主成分分析的自适应数据融合方法。该方法将变换后的主成分特征值作为权重,自适应地体现各主成分在代表原始数据信息量中的贡献率,使融合后的数据更能突出和反映各频率瞬时谱数据中的有效信息,实现了宽频带瞬时谱数据的降维与优化。实际资料的应用表明,本文的方法可快速提取和突出包含在大量瞬时谱数据集中的主要信息,清晰地刻画储层的几何形态和展布特征,提高资料解释的效率。In order to extract and highlight the primary feature information of reservoirs from a large number of common frequency attribute data generated by instantaneous spectral decomposition,an auto-adaptive data fusion method based on the principal components analysis(PCA) is presented in the paper.The method regards the eigenvalue of principal components after PCA as weighting,which adaptively reflects the contribution rate of each principal component in representing the amount of information of the original data.The method can highlight and reflect the effective information in each instantaneous spectral data more excellently.It achieves the dimensionality reduction and optimization of broad-band instantaneous data sets.The real data processing demonstrates that the method extracts fast and highlights the primary information contained in a large number of instantaneous spectral data set,clearly depicts the geometry and spatial distribution characteristics of reservoirs,and improves the efficiency of data interpretation.

关 键 词：主成分分析 自适应数据融合 瞬时谱分解 宽频带 

分 类 号：P631[天文地球—地质矿产勘探]