针对传统储层流体识别方法识别精度低、运算量大、过于依赖个人经验的缺点,提出基于密度聚类的K近邻法,根据待测层段测井数据的空间分布规律,将样本按相对密度聚类成数据簇,并利用K近邻投票获得各簇所属类别。将该方法应用在某油田奥陶系鹰山组碳酸盐岩储层识别中。结果表明,较之其他常用识别方法,该算法识别精度高,泛化性和鲁棒性强,在处理大数据分类问题时具有明显优势,且在识别常规方法难以识别的油水同层时取得了较好的效果,具有良好的应用前景,为利用数据挖掘方法解决油田勘探开发中的复杂问题提供了新思路。 Aiming at the shortcomings of traditional reservoir fluid identification methods, such as low recognition accuracy, large amount of computation and over-reliance on personal experience, a K-nearest neighbor method based on density clustering is proposed. According to the spatial distribution of logging data in the interval to be measured, Clustered into clusters, and use K neighbors to vote for each cluster belongs to the category. The method is applied to the identification of carbonate reservoirs of Yingshan Formation of Ordovician in an oil field. The results show that compared with other commonly used recognition methods, the proposed algorithm has high recognition accuracy, generalization and robustness. It has obvious advantages in dealing with big data classification problems, and has achieved the recognition of the same layer of oil and water which is difficult to be identified by conventional methods Good effect and good application prospect, which provides a new idea for using data mining methods to solve complex problems in oilfield exploration and development.