| 356 | 1 | 842 |
| 下载次数 | 被引频次 | 阅读次数 |
三维地质模型可以有效表征地下地层分布和地质构造，但受限于工程钻孔数据的稀疏性、不规则性、不平衡性，单一的机器学习方法在地质建模中往往难以达到理想精度。针对工程钻孔数据的特性，该文提出一种基于DBSCAN-SMOTEENN-RF联合算法改进的机器学习三维建模方法。首先根据地质资料调整算法参数以优化数据，进而创建研究区栅格单元地质属性模型，并与单一随机森林(RF)模型进行预测对比，最后进行不同数据处理方法的建模结果分析。实证结果表明，DBSCAN-SMOTEENN-RF联合算法能有效消除数据不平衡现象并提升建模效果，在数据量有限或质量不均的情况下，与单一RF模型在三维地质建模中的精度相比，该算法准确率、召回率、F1值和精确率分别提高8.38%、11.40%、10.12%、7.37%;在栅格单元地质属性模型的地层分布展示上，DBSCAN-SMOTEENN-RF模型的预测结果更符合勘察的地质情况。
Abstract：Three-dimensional geological models play a crucial role in accurately depicting the distribution of underground strata and geological structures.However, constrained by the sparsity, irregularity, and imbalance of engineering drill hole data, a single machine learning algorithm often struggles to achieve optimal accuracy in geological modeling.In light of the distinctive characteristics of engineering drill hole data, this paper proposes an improved machine learning-based 3D modeling method that employs a combined algorithm integrating DBSCAN(Density-Based Spatial Clustering of Applications with Noise),SMOTEENN(Synthetic Minority Over-sampling Technique combined with Edited Nearest Neighbors rule),and RF(Random Forest).First, the algorithm parameters are fine-tuned according to geological data, and the integrated algorithm is used to optimize the dataset.Subsequently, a grid cell geological attribute model of the study area is constructed using the RF model, and its predictions are compared with those of a standalone RF model.Finally, the results based on different data processing methods are analyzed.Validation results demonstrate that the DBSCAN-SMOTEENN-RF integrated algorithm effectively mitigates the data imbalance and enhances modeling performance.Under conditions of limited or uneven data quality, the proposed algorithm outperforms the single RF model in three-dimensional geological modeling, achieving improvements of 8.38%,11.40%,10.12% and 7.37% in accuracy, recall, F1 score and precision, respectively.Comparison of the constructed grid cell geological attribute models reveals that the model established based on DBSCAN-SMOTEENN-RF better reflects the geological conditions observed during exploration.
[1] 冉祥金.区域三维地质建模方法与建模系统研究[D].长春：吉林大学，2021.
[2] 潘炜，刘大安，钟辉亚，等.三维地质建模以及在边坡工程中的应用[J].岩石力学与工程学报，2004(4):597-602.
[3] 柴贺军，黄地龙，黄润秋，等.岩体结构三维可视化及其工程应用研究[J].岩土工程学报，2001(2):217-220.
[4] LEMON A M,JONES N L.Building solid models from boreholes and user-defined cross-sections[J].Computers & Geosciences,2003,29(5):547-555.
[5] 郭艳军，潘懋，王喆，等.基于钻孔数据和交叉折剖面约束的三维地层建模方法研究[J].地理与地理信息科学，2009,25(2):23-26.
[6] 陈麒玉，刘刚，吴冲龙，等.城市地质调查中知识驱动的多尺度三维地质体模型构建方法[J].地理与地理信息科学，2016,32(4):11-16.
[7] 邢延涛，李利军.三维地质体重构中空间数据插值方法的研究[J].计算机与数字工程，2006(12):45-47.
[8] CAUMON G,GRAY G,ANTOINE C,et al.Three-dimensional implicit stratigraphic model building from remote sensing data on tetrahedral meshes:theory and application to a regional model of La Popa Basin,NE Mexico[J].IEEE Transactions on Geoscience and Remote Sensing,2013,51(3):1613-1621.
[9] CALCAGNO P,CHILèS J P,COURRIOUX G,et al.Geological modelling from field data and geological knowledge:part I.modelling method coupling 3D potential-field interpolation and geological rules[J].Physics of the Earth and Planetary Interiors,2008,171(1):147-157.
[10] HILLIER M J,SCHETSELAAR E M,DE KEMP E A,et al.Three-dimensional modelling of geological surfaces using generalized interpolation with radial basis functions[J].Mathematical Geosciences,2014,46(8):931-953.
[11] 程曦文，李天斌，马春驰，等.一种多属性三维地质模型的快速建模方法及在高地应力隧道中的应用[J].工程地质学报，2023,31(3):959-967.
[12] ZENG Z Q,CAI Y C,WU J B.Local coupled markov chain model for simulating varied stratigraphy[J].Computers and Geotechnics,2024,167:106064.
[13] 周永章，左仁广，刘刚，等.数学地球科学跨越发展的十年：大数据、人工智能算法正在改变地质学[J].矿物岩石地球化学通报，2021,40(3):556-573.
[14] 张亦弛.基于机器学习的水电工程地质三维精细建模与混合现实可视化研究[D].天津：天津大学，2023.
[15] ELKATEB T,CHALATURNYK R,ROBERTSON P K.An overview of soil heterogeneity:quantification and implications on geotechnical field problems[J].Canadian Geotechnical Journal,2003,40(1):1-15.
[16] 郭甲腾，刘寅贺，韩英夫，等.基于机器学习的钻孔数据隐式三维地质建模方法[J].东北大学学报(自然科学版),2019,40(9):1337-1342.
[17] 徐学闯，张恒兵，符韶华，等.基于钻孔伪标签半监督深度学习的沈阳市浅表地层三维建模研究[J].地理与地理信息科学，2023,39(6):9-17.
[18] 马天寿，向国富，桂俊川，等.基于物理约束的分布式神经网络三维地应力预测模型[J].地球物理学报，2024,67(8):3211-3228.
[19] ORTIZ J M.Jef Caers:modeling uncertainty in the earth sciences[J].Mathematical Geosciences,2013,45(1):127-130.
[20] 梁栋.三维地质模型不确定性分析方法研究[D].武汉：中国地质大学，2022.
[21] 杜子纯，刘镇，明伟华，等.城市级三维地质建模的统一地层序列方法[J].岩土力学，2019,40(S1):259-266.
[22] 明镜.基于钻孔的三维地质模型快速构建及更新[J].地理与地理信息科学，2012,28(5):55-59.
[23] DEVI D,BISWAS S K,PURKAYASTHA B.Learning in presence of class imbalance and class overlapping by using one-class SVM and undersampling technique[J].Connection Science,2019,31(2):105-142.
[24] KARPATNE A,EBERT-UPHOFF I,RAVELA S,et al.Machine learning for the geosciences:challenges and opportunities[J].IEEE Transactions on Knowledge and Data Engineering,2019,31(8):1544-1554.
[25] ESHIMIAKHE D,LAWAL K.Application of K-means algorithm to Werner deconvolution solutions for depth and image estimations[J].Heliyon,2022,8(11):e11665.
[26] 杨楚龙，王怀秀，刘最亮.基于改进XGBoost的地震多属性地质构造识别方法[J].科学技术与工程，2023,23(29):12442-12450.
[27] 桂州，陈建国，王成彬.基于PCA-smote-随机森林的地质不平衡数据分类方法：以东天山地球化学数据为例[J].桂林理工大学学报，2017,37(4):587-593.
[28] WANG W T,WU Y L,TANG C Y,et al.Adaptive density-based spatial clustering of applications with noise(DBSCAN) according to data[C/OL]//2015 International Conference on Machine Learning and Cybernetics(ICMLC).[2024-04-22].https://ieeexplore.ieee.org/abstract/document/7340962.
[29] NEWALIYA N,SINGH Y.Multivariate hierarchical DBSCAN model for enhanced maritime data analytics[J].Data & Knowledge Engineering,2024,150:102282.
[30] HE H,GARCIA E A.Learning from imbalanced data[J].IEEE Transactions on Knowledge and Data Engineering,2009,21(9):1263-1284.
[31] MUNTASIR NISHAT M,FAISAL F,JAHAN RATUL I,et al.A comprehensive investigation of the performances of different machine learning classifiers with SMOTEENN oversampling technique and hyperparameter optimization for imbalanced heart failure dataset[J].Scientific Programming,2022,2022:e3649406.
[32] HO T K.The random subspace method for constructing decision forests[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1998,20(8):832-844.
基本信息:
中图分类号:P628
引用信息:
[1]王桂林,陈晓玲,岳佳豪,等.DBSCAN-SMOTEENN-RF联合算法及在三维地质建模中的应用[J].地理与地理信息科学,2025,41(03):19-26.
基金信息:
青海省科技计划项目(2024-ZJ-706); 重庆大学重庆市研究生科研创新项目(CYS240057)
2025-05-13
2025-05-13
2025-05-13