基于主成分分析的核Fisher判别方法在油水识别中的应用

Application of kernel Fisher method based on primary factor analysis to recognition problem between oil layer and water layer

摘要: 根据测井数据结构复杂和交集严重的特点,将主成分分析思想应用到剔除奇异点和寻找两类样本的交集中,并在交集中应用核Fisher判别方法,进行油水判别,弥补了Fisher线性判别方法的不足.通过将主成分分析和核Fisher判别方法这两种理论有机的结合起来,提高了利用测井数据识别油水层的鉴别能力,实际应用中证明了本方法的实用性和有效性.

Abstract: The idea of primary component analysis was applied to eliminating the singular point and selecting the intersection of raw log data sets according to the characteristics of raw log data. Then kernel Fisher method was used in the intersection, which remedy the shortcoming of linear differentiate methods. By combining the two method, primary component analysis and kernel Fisher, the differentiate capability was improved and the practicability is testified in application.