摘要
由于碳酸盐岩储层非均质性强、储集空间复杂多变、测井响应特征模糊,导致仅利用常规测井解释方法无法准确划分储集体类型。因此,将集成学习技术引入,以多种资料建立的测井响应模式为基础,提出一种结合Boosting和Bagging集成策略的改进方法应用于碳酸盐岩储集体类型划分中。采用AdaBoost.M2算法,以机器学习中的支持向量机、决策树、浅层神经网络为基分类器构建3个强学习器,并结合Bagging并行策略进行组合优化,得到储集体类型的最终划分结果。将该方法应用于塔河油田碳酸盐岩储层T615井组,结果表明:相较其他单分类器和基于同质基分类器的强分类器,本文方法的综合分类正确率最高,达92.3%,且对该井组的4类储集体的分类正确率均保持在90.0%左右,分类结果满足实际测井资料解释的精度要求,展现了集成学习技术在碳酸盐岩储集体类型划分中良好的应用效果。
Carbonate reservoirs are characteristic of high heterogeneity in substance and complexity in variable space, the fuzzy logging response is often obscured. It is impossible to identify accurately the types of reservoirs using conventional logging interpretation methods. Therefore, based on the logging response mode established by various data with the ensemble learning technology, an improved method combining Boosting and Bagging ensemble strategies was proposed for the classification of carbonate reservoir types. Using AdaBoost.M2 algorithm, three strong learners were built using base-classifiers of support vector machine, decision tree, and hidden layer neural network. Combined with the bagging parallel strategy for combination optimization, the final classification results of the reservoir types were obtained. The results of T615 well group carbonate reservoirs in the Tahe Oilfield show that compared with other single classifiers and strong classifiers based on homogeneous base-classifiers, the comprehensive classification accuracy rate of the proposed method is the highest, reaching 92.3%, and the classification accuracy rate of the four types of reservoirs in the well group is about 90.0%. The classification results can meet the accuracy requirements of the actual logging data interpretation, showing the good application effect of the ensemble learning technology in the classification of carbonate reservoir types.
作者
蓝茜茜
张逸伦
康志宏
徐嘉宏
孟顺
LAN Xi-xi;ZHANG Yi-lun;KANG Zhi-hong;XU Jia-hong;MENG Shun(School of Energy Resources,China University of Geosciences(Beijing),Beijing 100083,China;School of Earth and Space Sciences,Peking University,Beijing 100871,China)
出处
《科学技术与工程》
北大核心
2020年第18期7231-7238,共8页
Science Technology and Engineering
基金
国家科技重大专项(2017ZX05009-001)
中国地质调查局地质调查项目(DD20190085)。
