摘要
针对储层含油性识别过程的复杂性和不确定性,提出一种改进的蚁群聚类算法。将储层类别作为变量,以Jaccard系数衡量聚类结果与已知类结构的一致性,以类内样本与类中心的方差衡量类内的紧密度,利用改进的蚁群算法实现样本的最优划分。实验结果显示,该算法得到的聚类结果与已知的测井解释结论一致度高,类内的紧密程度高,对储层含油性识别具有良好的预测和检验能力。
Aiming to the complexity and uncertainty of oil-bearing of reservoir recognition,this paper proposes an improved ant colony clustering algorithm.It sets the category of reservoir as a variable,uses Jaccard index to measure the coincidence between clustering result and the given conclusion.The clustering compactness is evaluated by the virtue of the variance between the samples and the center in a cluster,and the optimal partition is achieved by means of the improved ant colony clustering algorithm.Experimental result indicates that the coincidence between the clustering result and the given logging interpretation is high and so is the coincidence in all clusters,so that the algorithm has preferable predictability and inspection capability
出处
《计算机工程》
CAS
CSCD
北大核心
2011年第13期178-180,共3页
Computer Engineering
基金
国家自然科学基金资助项目(70573101)
高等学校博士学科点专项科研基金资助项目(20070491011)
中国博士后基金资助项目(20090461293)
中央高校基本科研业务费基金资助项目(CUG090113)
中国地质大学(武汉)资源环境经济研究中心开放基金资助项目(2009B012)
关键词
软计算
蚁群算法
储层含油性识别
聚类
Jaccard系数
soft computing
ant colony algorithm
oil-bearing of reservoir recognition
clustering
Jaccard index
