| 摘要: |
| 对录井资料及测井资料与地层可钻性级值的关系进行分析，提出一种基于相关向量机算法的地层可钻性级值 预测的新方法。通过标准化钻速、测井声波时差、地层密度、泥质含量和地层深度进行学习训练相关向量机，建立地 层可钻性级值预测的相关向量机模塑。对准噶尔盆地部分井的地层可钻性级值进行预测的结果表明，该方法优于 BP神经网络方法，具有预测精度高、收敛速度快、推广能力强等优点。 |
| 关键词: 地层可钻性 相关向量机 录井资料 测井资料 预测模型 |
| DOI：10.3969/j.issn.1673-5005.2010.02.013 |
| 分类号:TE 249 |
| 基金项目:中国石化科技攻关项目(JP04014〉；山东省自然科学基金项目(ER2009FL029) |
|
| Formation drillability prediction based on relevance vector machine |
|
MA Hai1,WANG Yan-jiang1,HU Rui2,WEI Mao-an2
|
|
(1. College of Information and Control Engineering in China University of Petroleum, Dongying 257061, China;2. Drilling Technology Research Institute of Shengli Oilfield, Dongying 257017, China)
|
| Abstract: |
| The relation between log data, well-log data and formation drillability was analyzed, and a novel method for predicting formation drillability based on relevance vector machine ( RVM) was proposed. A prediction model for formation drillability was established using the data of normalized drilling rate, well-log acoustic velocity, formation density, shaliness and formation depth by training the RVM. The proposed method was applied to predict the formation drillability of some wells in Junggar Basin. The results show that the RVM method has higher prediction precision, faster convergence speed and better generalization effect than BP neural network approach. |
| Key words: formation drillability relevance vector machine log data weH-log data prediction model |
