摘要
模式识别应用于薄层油田勘探在国内外都是一个处于探索阶段的课题,这方面取得的进展相当少。本文阐述了对薄层砂岩进行厚度分类的方法,为物探人员的解释工作提供有用的参考数据。在专家经验缺乏、样本数量很少的情况下,根据已知的地震剖面数据,采用集群方法、近邻法则、留一交替法等手段,将薄层砂岩分类,得到77%以上的正确识别率,这表明将本文方法应用于薄层油田勘探是有效的。
The application of pattern recognition to the oil sheet exploration is a new research problem. In this field only a few progress has been made. In this paper, a method of sand streak classification is proposed. The sand streak is classified according to its thickness so that the exploration geophysicists could get some useful reference data. Using the clustering, the Nearest-Neighbour Algorithm, and the leave-one-out method, it gets a correct recognition rate of 77% under the lack of experts' experience and oil-well samples. The result demonstrates that this mothod is useful for oil sheet exploration.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
1994年第4期312-316,共5页
Pattern Recognition and Artificial Intelligence
关键词
模式识别
砂岩
分类
薄层砂岩
Pattern Recognition, Feature Extraction, Leave-One-Out Method, Classification of Sand Streak.
