论文详情
传统的测井解释规则库的获取需专业研究人员以手工方式进行,存在繁琐、耗时等缺点,且技术熟练程度直接影响到解释评价效果,为此,提出了一种基于XGBoost的测井解释规则库自动获取或建立测井解释专家规则库的方法,将多种物理信息和地质参数作为输入特征,储层类别作为输出标签,通过引入XGBoost算法,经过学习得出地质参数与储层类别之间的关系模型。利用该模型,可以自动预测储层类别,进而建立测井解释规则库。胜利油田盐家永安地区某砂砾岩油气藏的砂砾岩测井解释评价结果表明,与支持向量机(SVM)算法和梯度提升决策树(GBDT)算法相比,本文方法具有更高的准确率和更高的计算效率。研究区老井复查结果表明:与手工获取规则库方法相比,本文方法较完整地提取了研究区内的知识规则,提升了测井解释的准确率。
Logging data has traditionally been interpreted manually by professionals,which is a tedious and time-consuming task.Consequently,the interpretation accuracy has been directly affected by the technical proficiency of the professionals.In this study,a method for automatically establishing an expert rule base by means of the XGBoost algorithm was proposed.In this method,various physical data and geological parameters were used as input features,while reservoir types were the outputs.The XGBoost algorithm was able to establish a relationship between geological parameters and reservoir types,thereby ensuring that reservoir types could be predicted automatically and that criteria for log interpretation could be established.The method was applied to actual data from the Yanjia Yongan area of the Shengli oilfield.The results demonstrated that when compared with the support vector machine (SVM) and the original GBDT algorithm,the proposed XGBoost-based method can achieve higher accuracy and efficiency for the logging interpretation of a glutenite reservoir.Moreover,compared with the manual processing of log data,the proposed method can provide a more complete knowledge base,thus improving the accuracy of logging interpretation.
中国石油化工集团有限公司重大科技专项“中石化测井软件平台升级与应用子系统研发”(JPE19006)、中国石油天然气集团有限公司重大科技专项(ZD2019183008)共同资助。