基于KICA属性优化的支持向量机储层参数预测  

The research of reservoir parameters forecasting based on KICA and SVM

作　　者：王维强 WANG Wei-Qiang(Technology Innovation Center of Geothermal&Hot Dry Rock Exploration and Development,Ministry of Natural Resources,Shijiazhuang 050061,China;Institute of Hydrology and Environmental Geology,Chinese Academy of Geological Sciences,Shijiazhuang 050061,China)

机构地区：[1]自然资源部地热与干热岩勘查开发技术创新中心,河北石家庄050061 [2]中国地质科学院水文地质环境地质研究所,河北石家庄050061

出　　处：《物探与化探》2021年第4期990-997,共8页Geophysical and Geochemical Exploration

基　　金：中国地质调查局地质调查项目(DD20190128)。

摘　　要：为提高储层参数预测的精度,提出一种基于核独立分量分析(KICA)属性优化的支持向量机储层参数预测技术,KICA属性优化技术充分体现属性信息的非线性关系与高阶统计特性,提取出相互统计独立的反映地下储层参数的储层信息。支持向量机技术基于结构风险最小化原理,可解决小样本、高维与局部最小的非线性系统问题,二者有效的结合,能够将繁冗的地震属性空间,结合较少的井数据精确预测出储层的参数分布。通过模型及实际资料研究表明,本文储层参数预测方法的应用效果好,预测精度高。In order to improve the accuracy of prediction of reservoir parameters,the paper proposes the approach for reservoir parameters forecasting based on KICA and Support vector machine(SVM).The KICA attribute optimization technology reflects the non-linear relationship and high order statistical properties of the attributes,extract the reservoir information of mutual statistical independence which reflects the reservoir parameters of the subsurface.SVM technology based on structural risk minimization principle,which can solve problems of the nonlinear systems for the small sample,high dimensional and local minimum.KICA combined the SVM,which accurately predict the reservoir parameter distributions through the huge attribute space and less well data.Through the model and actual data,it shows that reservoir parameter prediction technology has good effect of application,and high prediction.

关 键 词：核独立分量分析 地震属性优化 支持向量机 储层参数 预测 

分 类 号：P631.4[天文地球—地质矿产勘探]