本文旨在研究基于Adaboost和Decision Tree算法的地层岩性预测方法,通过对气井的地层岩性实测数据进行分析,筛选出深度、地层电阻率等九种关键地球物理参数,利用上述机器学习算法构建气井地层岩性预测模型。在模型构建过程中,为解决Adaboost SAMME和Decision Tree算法参数选取和优化难点,利用交叉验证法筛选出最优参数组合。结果表明:Adaboost SAMME算法在岩性和地层岩性预测方面表现优异,准确率高达96%以上,相对而言,Decision Tree算法准确率稍低,为87%;模型预测准确率随训练集比例的增大而增加,原始数据随机化处理可以提高模型预测准确率;主成分分析(PCA)效果明显优于奇异值分解(SVD)。研究成果可为地下空间与能源工程钻井的地层岩性预测提供参考。
杨春曦
,
肖文梁
,
徐亚军
,
郝梓宇
,
鲍挺
. 基于Adaboost和DecisionTree的地层岩性预测研究[J]. 地下空间与工程学报, 2025
, 21(S2)
: 634
-642
.
DOI: 10.20174/j.JUSE.2025.S2.12
This paper aims to investigate a method for predicting formation and lithology based on the Adaboost and Decision Tree algorithms. By analyzing the measured formation and lithology data from gas wells, nine key geophysical parameters, including depth and formation resistivity, were selected. These machine learning algorithms were employed to construct a prediction model for the formation and lithology of a gas well. During the model construction process, cross-validation was used to select the optimal parameter combination to address the challenges of parameter selection and optimization in the Adaboost SAMME and Decision Tree algorithms. The results indicate that: The Adaboost SAMME algorithm performs excellently in lithology and stratigraphic lithology prediction, with an accuracy exceeding 96%. In contrast, the Decision Tree algorithm has a slightly lower accuracy of 87%. The prediction accuracy of the model increases with the proportion of the training set; randomization of the original data can improve the model's prediction accuracy; and principal component analysis (PCA) is significantly more effective than singular value decomposition (SVD). The findings of this study provide an effective and rapid response method for predicting stratigraphic lithology in underground space and energy engineering drilling.
[1] Akinyokun O C,Enikanselu P A,Adeyemo A B,et al.Well log interpretation model for the determination of kithology and fluid contents[J].The Pacific Journal of Science and Technology,2009:507-517.
[2] Delfiner P C,Peyret O,Serra O.Automatic determination of lithology from well logs[J].SPE Formation Evaluation,1987,2(3):303-310.
[3] 李瑞,杜奉屏,肖崇礼.用ISODATA统计法识别碳酸盐岩气水层[J].石油物探,1990(1):99-113,39.
[4] 雍世和,陈钢花.应用Bayes逐步判别分析自动确定岩性[J].石油物探,1990(2):68-77.
[5] Xu H,Yan J,Feng G,et al.Rock layer classification and identification in ground-penetrating radar via machine learning[J].Remote Sensing,2024,16(8):1310.
[6] 庞志超,肖华,毛晨飞,等.准噶尔盆地南缘地区含膏质地层岩性特征及测井识别方法[J].吉林大学学报(地球科学版),2024,54(4):1419-1431.
[7] 靳九龙,杨斌,唐生寿,等.西湖凹陷N构造花港组厚层砂泥岩地层测井岩性识别方法[J].石化技术,2024,31(8):218-220.
[8] 段忠义,肖昆,杨亚新,等.松辽盆地砂岩型铀矿钻孔岩性的测井识别[J].地球物理学进展,2023,38(6):2490-2501.
[9] Freund Y,Schapire R E.A Decision-theoretic generalization of on-line learning and an application to boosting[J].Journal of Computer and System Sciences,1997,55(1):119-139.
[10] Quinlan J R.Generating production rules from decision trees[A]//Proceedings of the 10th International Joint Conference on Artificial Intelligence[C].San Francisco,CA,USA:Morgan Kaufmann Publishers Inc.,1987:304-307.
[11] Xie Y,Zhu C,Zhou W,et al.Evaluation of machine learning methods for formation lithology identification:A comparison of tuning processes and model performances[J].Journal of Petroleum Science and Engineering,2018,160:182-193.
