摘要
岩性岩相识别与划分是地层评价和油藏精细描述的一项重要工作。在岩性岩相分类中,基于Scikit-learn机器学习框架,采用岩心观察描述和薄片分析数据划分了鄂尔多斯盆地合水地区长7地层的岩性岩相,形成了机器学习样本库训练集和测试集;运用Python编程软件,编写基于KNN(K-近邻算法)的机器学习模块,对训练集中的数据进行学习,形成预测模型,用测试集对模型进行测试评价,测试结果显示KNN模型分类准确率为89.5%,总体预测效果较好,为后续储层三维精细建模提供了技术支持。
Identification and classification of lithology and lithofacies is important for formation evaluation and fine reservoir description.In this study,first,based on the Scikit-learn framework,the lithology and lithofacies of the Chang 7 Member in the Heshui area the Ordos Basin were divided according to core observation and slice analysis;then a training set and a testing set were established as a machine learning sample database;and finally using Python software,a KNN(K-near neighbor)machine learning module was built,which developed a formation forecast model after learning the training data.The KNN model was evaluated by the testing data to classification accuracy of 89.5%.Good forecast effect means a technical support for building a fine 3D reservoir model in the Heshui area.
作者
陈玉林
李戈理
杨智新
肖飞
车锐媚
陈彦竹
CHEN Yulin;LI Geli;YANG Zhixin;XIAO Fei;CHE Ruimei;CHEN Yanzhu(China Petroleum Logging CO. LTD., Xi’an, Shaanxi 710077, China;Shaanxi LNG Investment & Development CO. LTD., Yangling, Shaanxi 712100, China)
出处
《测井技术》
CAS
2020年第2期182-185,共4页
Well Logging Technology
基金
中国石油集团公司重大专项(2016D-3803)。
关键词
测井评价
岩性岩相
KNN
机器学习
合水地区
log evaluation
lithology and lithofacies
KNN
machine learning
Heshui area
作者简介
第一作者:陈玉林,男,1985年生,硕士,工程师,从事测井资料解释与评价研究,E-mail:chenyl123@cnpc.com.cn。
