| 本文已被：浏览次 下载次 |
码上扫一扫！ |
|
|
| 基于时域统计特征的井眼碰撞识别方法 |
|
刘刚1,张家林1, 刘闯2,刘华亮3,于长广4
|
|
(1.中国石油大学石油工程学院,山东青岛 266580;2.中国电波传播研究所,河南新乡 453000;3.中石化胜利钻井工程技术公司定向井公司,山东东营 257064;4.中海油能源发展股份有限公司工程技术分公司,天津 300452)
|
|
| 摘要: |
| 相同钻井条件下,钻头钻遇水泥环、套管的振动信号与钻进地层的振动信号的时域特征不同。提出一种基于时域统计特征的井眼碰撞识别方法,通过识别振动信号的特征差异,实现对钻头碰撞邻井井眼的及时预警。首先在时域上提取信号的分散特征和形状分布特征,然后利用主成分分析法对时域特征进行降维处理,进而得到主要特征成分,最后利用支持向量机对特征进行分类训练,以完成对钻进地层和钻遇水泥环、套管的自动识别。通过南海某钻井平台现场作业数据检验,验证了该方法的可行性,表明该方法可以加强海上丛式井作业的安全性。 |
| 关键词: 井眼碰撞 时域特征 主成分分析法 支持向量机 |
| DOI：10.3969/j.issn.1673-5005.2016.02.008 |
| 分类号::TE 52 |
| 文献标识码:A |
| 基金项目:“十二五”国家科技重大专项(2011ZX05057-002-006,1ZX05024-002-010);山东省自然科学基金项目(ZR2014EEQ021);中央高校基本科研业务费专项(14CX02167A) |
|
| A borehole collision recognition method based on the time statistical characteristics |
|
LIU Gang1, ZHANG Jialin1, LIU Chuang2, LIU Hualiang3, YU Changguang4
|
|
(1.School of Petroleum Engineering in China University of Petroleum, Qingdao 266580, China;2.China Research Institute of Radio Propagation, Xinxiang 453000, China;3.Shengli Drilling Engineering and Technology Company, SINOPEC, Dongying 257064, China;4.CNOOC Energy Technology Drilling ＆ Production Corporation, Tianjin 300452, China)
|
| Abstract: |
| During well drilling, special time-domain features of drilling bit vibration signals will be produced when drilling at cement sheath and casing, which are different with that when drilling through rock formations. A new method for the recognition of borehole collision was proposed by analyzing the vibration signals produced when the drill bit collided with the cement sheath and casing of nearby wells. Firstly, the dispersion and shape distribution characteristics of the vibration signals were extracted, then the principal component analysis(PCA) was conducted to obtain the principal features of the signals. Finally a support vector machine (SVM) was trained with sampled signals to establish a model, which could be used to identify the borehole collision automatically by distinguishing the principal features of different vibration signals. This method has been effectively verified through real drilling data analysis from offshore cluster wells. |
| Key words: borehole collision time-domain feature principal component analysis support vector machine |
