数据挖掘方法在测井岩性识别中的应用

Application of data mining method in lithology identification using well log

 李政宏 刘永福 张立强 赵海涛 陈曦 李昊东

中国石油大学（华东）地球科学与技术学院，山东 青岛 266580 中国石油塔里木油田公司勘探开发研究院，新疆 库尔勒 841000 帝国理工学院地球科学与工程系，伦敦 SW72AZ 中国石油大港油田公司第二采油厂，天津 061103）

School of Geosciences, China University of Petroleum, Qingdao 266580, China Research Institute of Exploration and Development, Tarim Oilfield Company, PetroChina, Korla 841000, China Imperial College London, London SW72AZ, U.K. No.2 Oil Production Plant, Dagang Oilfield Company, PetroChina, Tianjin 061103, China

 测井岩性识别是油气藏勘探开发的重要基础工作。随着计算机技术的发展，数据挖掘方法越来越多地应用于岩性识别以提高预测准确性。数据挖掘方法可归纳为多元统计算法和智能性算法两大类，其中多元统计算法包括主成分分析、判别分析，智能性算法有神经网络、决策树、支持向量机。目前多元统计算法在测井岩性识别中应用广泛，智能性算法的应用尚处于发展阶段。基于大量文献调研的成果，概述了多元统计算法的原理及应用现状，重点梳理智能性算法的理论和优势，提出在应用智能性算法时需要将测井数据预处理，包括测井参数选择、测井数据归一化和降维。在此基础上，通过实例验证了智能性算法的应用效果，认为这是测井岩性识别领域今后的发展方向。

 Logging lithology identification is an important foundation work for oil and gas reservoir exploration and development. With the development of computer technology, data mining methods are increasingly applied to lithology identification to improve prediction accuracy. Data mining methods can be summarized into two categories: multivariate statistical algorithms and intelligent algorithms. Multivariate statistical algorithm includes principal component analysis and discriminant analysis. Intelligent algorithm includes neural networks, decision trees, and support vector machines. At present, multivariate statistical algorithms are widely used in logging lithology identification, and the application of intelligent algorithms is still in the development stage. Based on the results of a large number of literature research, the principle and application status of multivariate statistical algorithms are summarized, and the theory and advantages of intelligent algorithms are summarized. It is proposed that the logging data needs to be preprocessed when applying the intelligent algorithm, including logging parameter selection, logging data normalization and logging data dimensionality reduction. On this basis, the application of intelligent algorithm is verified by examples as the future development direction of logging lithology identification.

10.6056/dkyqt201906007