特征聚类在油田测试方案优化中的研究  

Research on optimization of oilfield test scheme based on characteristic clustering

作　　者：李洪奇[1,2] 张艳丽 杨景海[3] 朱丽萍 赵艳红[1,2] 裴建亚[3] 

机构地区：[1]中国石油大学地球物理与信息工程学院,北京102249 [2]中国石油大学油气数据挖掘北京市重点实验室,北京102249 [3]大庆油田测试技术服务分公司,黑龙江大庆163000

出　　处：《计算机工程与应用》2017年第21期214-218,共5页Computer Engineering and Applications

摘　　要：针对油田注产剖面动态测试在选井上没有一个合适参考标准的问题,提出利用基于特征的聚类将油井按照生产状况进行先聚类再分类的方法。首先对油井生产时间序列数据选择处理,然后提取序列特征进行聚类,并把聚类结果划分等级,最后制定油田测试优化方案。实验结果表明,提取的时间序列特征能很好地表征油井生产波动情况,获得较好地分类效果,对指导油田测试有重要意义。Note the oilfield production cross section in the dynamic test in the well election does not have a suitable reference standard issue, the paper proposes a well production division method in accordance with the feature-based clustering and the classification. First, the method selects and processes the well production time series data. And then the characteristic sets are extracted for clustering from time series data. In the end the clustering results develop the oilfield test optimization. Experimental result shows that the characteristics extracted from time series can be a good representation of the fluctuations in oil production, obtain better classification results and be important for guiding the oilfield test.

关 键 词：特征 聚类和分类 时间序列 油田测试优化 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]