摘要
对新疆油田公司的测井数据,利用核Fisher判别分析法(KFDA)判别油层水层。首先通过非线性映射(由核函数隐含定义)将样本映射到特征空间,然后在特征空间中用fisher判别分析(FDA)进行分类。实验结果表明,KFDA方法的预测准确率达92.9%,高于用Fisher判别分析法及人工神经网络(ANN)、支持向量机(SVM)方法进行判别的准确率。
This paper determines the oil layer and water layer with the well logging data of Xinjiang Oilfield Company by using Kernel Fisher Discriminant Analysis (KFDA). Firstly, it maps the sample to the feature space through the nonlinear mapping (by the kernel function definition), and then uses the FDA for classification in feature space. Experimental results show that the prediction accuracy rate of KFDA method is 92.9%, higher than with Fisher discriminant analysis method, artificial neural network (ANN) and the support vector machine (SVM) method.
出处
《价值工程》
2015年第11期172-174,共3页
Value Engineering
基金
陕西省自然科学基础研究计划项目(2014JM1032)
陕西省教育厅科学研究计划项目(2013JK1125)
国家级大学生创新训练计划项目(201310722009
1814
2013027)
关键词
油层
水层
KFDA
预测
oil layer
water layer
KFDA
forecast