中文摘要：

通过分析地震属性与储层预测对象的相关关系,提出了一种基于地震属性相关主成分分析的油气储层预测方法,该方法通过降维获取地震属性综合变量,该综合变量既能代表地震属性变异或数据的空间结构,又能保证地震属性与储层参数具有最大的相关性,综合变量个数由卡纳尔准则确定,理论试算及实际应用表明该方法具有良好的去噪作用,能形成较可信的地震属性与储层参数的关系,具有较高的预测精度,本文方法优于基于地震属性主成分分析的储层预测方法。

英文摘要：

The relation between seismic attributes and predicted object is not considered in extracting principal components by seismic attribute principal component analysis,so an improved principal component is proposed.It demand the principal components must represent the structure of data space and keep max correlation between seismic attributes and predicted object,the relation is gotten by optimized algorithm.The numbers of seismic attribute principal compents is gotten from Kanal rule.This method can eliminate the noise and get correct relation between seismic attributes and predicted objects.The computation is made on theory data and practical data,the result shows this method is better than seismic attribute principal component alalysis.