一种新型分类器在用测井资料预测油气中的应用

提出了一种在碳酸岩地层中划分油气层的新方法。此方法利用广义线性分类器和最近邻分类法则相结合的模型，叙述了分类器设计原理和特征选择的方法。该方法灵活并有对不同地质条件的自适应能力。试验证明这种分类方法效果显著，分类错误率低于10％，并给出了上扬子地区实际测井资料处理的结果。

New Classifier to Predict Oil and Gas Reservoir by Well Logging Data

A new approach to carbonate analysis, which uses Pattern Recognition technique, is presented in this paper. The proposed method combines the Decision Tree Classifier and the Nearest Neighbor Decision Rule .In each step of the tree classifier, a generalized linear discriminate function is used. The design of the function and selection of features are also described. The method is very flexible and has self adaptability to fit lots of geological condition. Experiments show that the approach is very effective with lower error rate of classification(<10%).The results of processed real oil well logs from upper Yangzi area are also given.