聚类分析在井下燃爆判别中的应用  

Application of Clustering Analysis in Underground Explosion Discrimination

作　　者：崔真[1] 

机构地区：[1]中国石化胜利石油管理局钻井工艺研究院,山东东营257017

出　　处：《石油钻探技术》2010年第6期20-24,共5页Petroleum Drilling Techniques

基　　金：国家高技术研究发展计划("863"计划)项目"超深井钻井技术研究"(编号:2006AA06A109)部分研究内容

摘　　要：聚类分析是一种在安全分析领域应用广泛的多元统计方法,其在安全评价、监测、分析中使用效果显著。将井下燃爆类型判别抽象为一个数学分类问题,应用聚类分析的方法来进行分类,对聚类分析在井下燃爆判别中的应用方法进行了研究,提出了一种现场测试数据处理的标准化原则,并应用系统聚类分析方法对川东北地区某井的现场测试数据进行了分析计算。应用最短距离法、最长距离法、中间距离法、重心法、类平均法、离差平方和法多种聚类准则得出的结果高度一致,并与现场判别软件的判别结果一致,证实了该方法的有效性。该方法减少了数据冗余,原理清晰,计算简单,其结论客观,符合现场情况。Clustering analysis is a multivariate statistical method widely used in the field of security analysis in recent years.Its effect is remarkable in safety evaluation,monitoring and analysis.The type of explosion discrimination was regarded as a mathematical classification,and clustering analysis was used to classify the explosion with statistical ways.This paper proposed a standardized principle of processing field test data,and applied system clustering method to analyze and calculate the field data of a well in Northeast Sichuan area.The results are consistent with the multi-clustering principle,including shortest-distance method,longest-distance method,mid-distance method,barycenter method,deviation average method and sum of deviation square method,and consistent with the results of distinguishing software used in wellsite.The effectiveness of this method was proved,and it reduced the data redundancy.The principle is clear,and the computation is simple.The conclusion is in accordance with the field condition.

关 键 词：气体钻井 井下燃爆 聚类分析 数据处理 

分 类 号：TE242.6[石油与天然气工程—油气井工程]