TPE-SVM Model and SHAP Analysis to Identify Pb-Zn Deposit Types Based on Sphalerite Trace Elements
-
摘要: 为了解闪锌矿微量元素特征对不同成因矿床类型是否能够进行有效判别，系统收集了全球典型的沉积喷流型（SEDEX）、密西西比河谷型（MVT）、火山块状硫化物型（VMS）、矽卡岩型（skarn）和浅成低温热液型（epithermal）铅锌矿床中3 117条闪锌矿的12种微量元素含量数据（Mn、Fe、Co、Cu、Ga、Ge、Ag、Cd、In、Sn、Sb、Pb），使用基于Tree-structured Parzen Estimator（TPE）优化的支持向量机机器学习算法建立了闪锌矿微量元素分类模型，并使用SHAP（SHapley Additive exPlanations）方法进行特征重要性分析.结果表明，经优化的TPE-SVM模型在测试集上展现出优异的分类能力，准确率、召回率和F1值均超过0.97.通过SHAP解释发现闪锌矿中Mn、Ge、Co为矿床成因类型判别三大关键元素.本文建立的闪锌矿微量元素判别指标体系，不仅为矿床成因鉴定提供了新的技术手段，更可为复合成矿系统解析、隐伏矿体预测等复杂地质问题提供创新解决方案.Abstract: This study demonstrates the efficacy of machine learning algorithms in classifying genetic types of Pb-Zn deposits through trace elements in sphalerite. It compiled a comprehensive trace element dataset comprising 3 117 sphalerite samples from 109 globally representative Pb-Zn deposits including MVT, VMS, SEDEX, skarn, and epithermal deposits. Twelve trace elements (Mn, Fe, Co, Cu, Ga, Ge, Ag, Cd, In, Sn, Sb, Pb) were systematically analyzed to develop a Tree-structured Parzen Estimator (TPE)-optimized Support Vector Machine (SVM) classification model. The model demonstrated exceptional discriminative performance on test datasets, achieving accuracy, recall, and F1-score values exceeding 0.97. SHAP (SHapley Additive exPlanations) interpretability analysis revealed Mn, Ge, and Co as critical discriminators among deposit types, providing quantitative insights into elemental controls on genetic classification. The discriminant index system of trace elements in sphalerite established in this paper not only provides a new technical means for the identification of ore genesis, but also provides innovative solutions for complex geological problems such as the analysis of composite metallogenic system and the prediction of concealed ore bodies.
-
Key words:
- sphalerite /
- trace elements /
- machine learning /
- TPE optimization algorithm /
- SHAP algorithm /
- ore deposits
-
图 1 闪锌矿数据集所涵盖的矿床位置分布
SEDEX.沉积喷流型；MVT.密西西比河谷型；VMS.火山块状硫化物型；skarn.矽卡岩型；epithermal.浅成低温热液型；下同
Fig. 1. Deposit location distribution map of the sphalerite dataset
图 3 闪锌矿微量元素对数变换后数据统计直方图及正态分布拟合曲线
纵坐标表示数据量
Fig. 3. Statistical histograms and fitting curves of logarithmic transformation of trace elements in sphalerite
图 6 准确率、精确率、召回率和F1分数指标计算图
定义：TP为真阳性，TN为真阴性，FP为假阳性，FN为假阴性
Fig. 6. Diagram illustrating the calculation of metrics: accuracy, precision, recall and F1-score
图 8 分类模型在测试集的混淆矩阵
Fig. 8. The confusion matrix of the classification model in the test set
图 10 各元素对分类预测的特征重要性排序图（a）和TPE-SVM模型中每个类别的特征重要性（b-f）
横坐标为每个特征对每个样本数据的平均SHAP值大小，颜色对应元素浓度大小
Fig. 10. The feature importance ranking diagram of each element for classification prediction(a)and feature importance for each category in the TPE-SVM model (b-f)
图 11 富乐矿床模型预测力图
Fig. 11. Force plots showing the composite model predictions for the Fule deposit
表 1 模型的性能评价指标
Table 1. Performance evaluation index of model
模型 矿床类型 精确率 准确率 召回率 F1分数 AUC TPE-SVM MVT 0.98 1.00 1.00 1.00 1.00 SEDEX 0.98 0.97 0.98 1.00 VMS 0.99 0.99 0.99 1.00 浅成低温热液型 0.98 0.99 0.99 1.00 矽卡岩型 0.97 0.97 0.97 1.00 
下载: 导出CSV
-
Aggarwal, C. C., 2016. An Introduction to Outlier Analysis. Springer International Publishing, Cham, 1-34. https://doi.org/10.1007/978-3-319-47578-3_1 Bauer, M. E., Burisch, M., Ostendorf, J., et al., 2019. Trace Element Geochemistry of Sphalerite in Contrasting Hydrothermal Fluid Systems of the Freiberg District, Germany: Insights from LA-ICP-MS Analysis, Near-Infrared Light Microthermometry of Sphalerite-Hosted Fluid Inclusions, and Sulfur Isotope Geochemistry. Mineralium Deposita, 54(2): 237-262. https://doi.org/10.1007/s00126-018-0850-0 Bédard, É., De Bronac de Vazelhes, V., Beaudoin, G., 2022. Performance of Predictive Supervised Classification Models of Trace Elements in Magnetite for Mineral Exploration. Journal of Geochemical Exploration, 236: 106959. https://doi.org/10.1016/j.gexplo.2022.106959 Belissont, R., Muñoz, M., Boiron, M. C., et al., 2016. Distribution and Oxidation State of Ge, Cu and Fe in Sphalerite by μ-XRF and K-Edge μ-XANES: Insights into Ge Incorporation, Partitioning and Isotopic Fractionation. Geochimica et Cosmochimica Acta, 177: 298-314. https://doi.org/10.1016/j.gca.2016.01.001 Cai, Y. W., Qiu, K. F., Petrelli, M., et al., 2024. The Application of "Transfer Learning" in Optical Microscopy: The Petrographic Classification of Opaque Minerals. American Mineralogist, 109(12): 2060-2072. https://doi.org/10.2138/am-2023-9092 Caleb, C. J., Gysi, A. P., Monecke, T., et al., 2023. Experimental Study of Apatite-Fluid Interaction and Partitioning of Rare Earth Elements at 150 and 250 ℃. American Mineralogist, 108(8): 1409-1420. https://doi.org/10.2138/am-2022-8589 Chen, J. Y., Shen, H. J., Yan, W. Y., 2023. LA-ICP-MS Trace Element Geochemistry of Sphalerite: Metallogenic Constraints on the Langyaquan Pb-Zn Deposit in the Middle Tianshan Metallogenic Belt. Mineral Exploration, 14(3): 377-391 (in Chinese with English abstract). Cook, N. J., Ciobanu, C. L., Pring, A., et al., 2009. Trace and Minor Elements in Sphalerite: A LA-ICPMS Study. Geochimica et Cosmochimica Acta, 73(16): 4761-4791. https://doi.org/10.1016/j.gca.2009.05.045 Dong, S. N., Wang, D., Ma, G. T., et al., 2024. Application of Machine Learning to Predict Types of Pb-Zn Deposits by Using Trace Elemental Characteristics of Sphalerite. Journal of Chengdu University of Technology (Science & Technology Edition), 51(4): 614-629 (in Chinese with English abstract). Dong, Y. L., Zhang, Z. J., 2024. Deep Forest Modeling: An Interpretable Deep Learning Method for Mineral Prospectivity Mapping. Journal of Geophysical Research: Machine Learning and Computation, 1(4): e2024JH000311. https://doi.org/10.1029/2024JH000311 Frenzel, M., Hirsch, T., Gutzmer, J., 2016. Gallium, Germanium, Indium, and Other Trace and Minor Elements in Sphalerite as a Function of Deposit Type—A Meta-Analysis. Ore Geology Reviews, 76: 52-78. https://doi.org/10.1016/j.oregeorev.2015.12.017 Gisbert, G., Tornos, F., Losantos, E., et al., 2021. Vectors to Ore in Replacive Volcanogenic Massive Sulfide (VMS) Deposits of the Northern Iberian Pyrite Belt: Mineral Zoning, Whole Rock Geochemistry, and Application of Portable X-Ray Fluorescence. Solid Earth, 12(8): 1931-1966. https://doi.org/10.5194/se-12-1931-2021 Guilbert, J. M., Park, C. F., 1986. The Geology of Ore Deposits. Freeman, New York. He, L., Liang, T., Wang, D. H., et al., 2024. Skarn Formation and Zn-Cu Mineralization in the Dachang Sn Polymetallic Ore Field, Guangxi: Insights from Skarn Rock Assemblage and Geochemistry. Minerals, 14(2): 193. https://doi.org/10.3390/min14020193 Hou, L. L., Wu, S., Yi, J. Z., et al., 2024. Discriminating Deposit Types Using Chlorite Trace Elements Based on Machine Learning. Earth Science, 49(12): 4303-4317 (in Chinese with English abstract). Huang, H. X., Li, R. X., Xiong, F. Y., et al., 2020. A Method to Probe the Pore-Throat Structure of Tight Reservoirs Based on Low-Field NMR: Insights from a Cylindrical Pore Model. Marine and Petroleum Geology, 117: 104344. https://doi.org/10.1016/j.marpetgeo.2020.104344 Huang, X. H., Li, Z. H., Deng, T., et al., 2023. Uranium Potential Evaluation of Zhuguangshan Granitic Pluton in South China Based on Machine Learning. Earth Science, 48(12): 4427-4440 (in Chinese with English abstract). Jin, L. Y., Qin, K. Z., Li, G. M., et al., 2015. Trace Element Distribution in Sulfides from the Chalukou Porphyry Mo-Vein-Type Zn-Pb System, Northern Great Xing'an Range, China: Implications for Metal Source and Ore Exploration. Acta Petrologica Sinica, 31(8): 2417-2434 (in Chinese with English abstract). Keith, M., Haase, K. M., Schwarz-Schampera, U., et al., 2014. Effects of Temperature, Sulfur, and Oxygen Fugacity on the Composition of Sphalerite from Submarine Hydrothermal Vents. Geology, 42(8): 699-702. https://doi.org/10.1130/g35655.1 Li, X. M., Zhang, Y. X., Li, Z. K., et al., 2023. Discrimination of Pb-Zn Deposit Types Using Sphalerite Geochemistry: New Insights from Machine Learning Algorithm. Geoscience Frontiers, 14(4): 101580. https://doi.org/10.1016/j.gsf.2023.101580 Liu, P., Mao, J. W., Jian, W., et al., 2020. Fluid Mixing Leads to Main-Stage Cassiterite Precipitation at the Xiling Sn Polymetallic Deposit, SE China: Evidence from Fluid Inclusions and Multiple Stable Isotopes (H-O-S). Mineralium Deposita, 55(6): 1233-1246. https://doi.org/10.1007/s00126-019-00933-0 Liu, W. H., Mei, Y., Etschmann, B., et al., 2023. Germanium Speciation in Experimental and Natural Sphalerite: Implications for Critical Metal Enrichment in Hydrothermal Zn-Pb Ores. Geochimica et Cosmochimica Acta, 342: 198-214. https://doi.org/10.1016/j.gca.2022.11.031 Lundberg, S., Lee, S. I., 2017. A Unified Approach to Interpreting Model Predictions. Computer Science, 1-10. https://doi.org/10.48550/arXiv.1705.07874 Martin, A. J., McDonald, I., McFall, K. A., et al., 2021. Low-Temperature Silica-Rich Gold Mineralization in Mafic VMS Systems: Evidence from the Troodos Ophiolite, Cyprus. Mineralium Deposita, 56(4): 805-822. https://doi.org/10.1007/s00126-020-01007-2 Nathwani, C. L., Wilkinson, J. J., Fry, G., et al., 2022. Machine Learning for Geochemical Exploration: Classifying Metallogenic Fertility in Arc Magmas and Insights into Porphyry Copper Deposit Formation. Mineralium Deposita, 57(7): 1143-1166. https://doi.org/10.1007/s00126-021-01086-9 Niu, P. P., Muñoz, M., Mathon, O., et al., 2024. Mechanism of Germanium Enrichment in the World-Class Huize MVT Pb-Zn Deposit, Southwestern China. Mineralium Deposita, 59(5): 995-1016. https://doi.org/10.1007/s00126-023-01242-3 Qiu, K. F., Zhou, T., Chew, D., et al., 2024. Apatite Trace Element Composition as an Indicator of Ore Deposit Types: A Machine Learning Approach. American Mineralogist, 109(2): 303-314. https://doi.org/10.2138/am-2022-8805 Rajabpour, S., Hassanpour, S., Jiang, S. Y., 2023. Physicochemical Evolution and Mechanism of a Skarn System: Insights from the World-Class Mazraeh Cu Deposit, NW Iran. GSA Bulletin, 136(1-2): 351-370. https://doi.org/10.1130/B36854.1 Reich, M., Deditius, A., Chryssoulis, S., et al., 2013. Pyrite as a Record of Hydrothermal Fluid Evolution in a Porphyry Copper System: A SIMS/EMPA Trace Element Study. Geochimica et Cosmochimica Acta, 104: 42-62. https://doi.org/10.1016/j.gca.2012.11.006 Ren, T., Zhou, J. X., Wang, D., et al., 2019. Trace Elemental and S-Pb Isotopic Geochemistry of the Fule Pb-Zn Deposit, NE Yunnan Province. Acta Petrologica Sinica, 35(11): 3493-3505 (in Chinese with English abstract). doi: 10.18654/1000-0569/2019.11.15 Sangster, D. F., 2018. Toward an Integrated Genetic Model for Vent-Distal SEDEX Deposits. Mineralium Deposita, 53(4): 509-527. https://doi.org/10.1007/s00126-017-0755-3 Sangster, D. F., 2020. Evidence for Lateral Fluid Flow in Vent-Distal SEDEX Zn-Pb Deposits and Districts. Mineralium Deposita, 55(3): 399-407. https://doi.org/10.1007/s00126-019-00931-2 Sihombing, F. M. H., Palin, R. M., Hughes, H. S. R., et al., 2024. Improved Mineral Prospectivity Mapping Using Graph Neural Networks. Ore Geology Reviews, 172: 106215. https://doi.org/10.1016/j.oregeorev.2024.106215 Song, Y. C., Yang, Z. M., Zhuang, L. L., 2023. Enrichment of Mississippi Valley-Type (MVT) Deposits in the Tethyan Domain Linked to Organic Matter-Rich Sediments. Science China Earth Sciences, 66(12): 2853-2870. https://doi.org/10.1007/s11430-023-1195-5 Straumal, B., Kilmametov, A., Korneva, A., et al., 2021. The Enrichment of (Cu, Sn) Solid Solution Driven by High-Pressure Torsion. Crystals, 11(7): 766. https://doi.org/10.3390/cryst11070766 Sun, G. T., Zhou, J. X., 2022. Application of Machine Learning Algorithms to Classification of Pb-Zn Deposit Types Using LA-ICP-MS Data of Sphalerite. Minerals, 12(10): 1293. https://doi.org/10.3390/min12101293 Tăma, C. G., Andrii, M. P., Kovács, R., et al., 2021. Sphalerite Composition in Low- and Intermediate-Sulfidation Epithermal Ore Bodies from the Roia Montană Au-Ag Ore Deposit, Apuseni Mountains, Romania. Minerals, 11(6): 634. https://doi.org/10.3390/min11060634 Tan, R. C., Shao, Y. J., Brzozowski, M. J., et al., 2024. Development of a Machine Learning Model to Classify Mineral Deposits Using Sphalerite Chemistry and Mineral Assemblages. Ore Geology Reviews, 169: 106076. https://doi.org/10.1016/j.oregeorev.2024.106076 Torró, L., Millán-Nuñez, A. J., Benites, D., et al., 2023. Germanium- and Gallium-Rich Sphalerite in Mississippi Valley-Type Deposits: The San Vicente District and the Shalipayco Deposit, Peru. Mineralium Deposita, 58(5): 853-880. https://doi.org/10.1007/s00126-023-01160-4 Wang, C. Y., Li, J. F., Wang, K. Y., et al., 2018. Geology, Fluid Inclusion, and Stable Isotope Study of the Skarn-Related Pb-Zn (Cu-Fe-Sn) Polymetallic Deposits in the Southern Great Xing'an Range, China: Implications for Deposit Type and Metallogenesis. Arabian Journal of Geosciences, 11(5): 88. https://doi.org/10.1007/s12517-018-3417-6 Wang, Y., Qiu, K. F., Müller, A., et al., 2021. Machine Learning Prediction of Quartz Forming-Environments. Journal of Geophysical Research: Solid Earth, 126(8): e2021JB021925. https://doi.org/10.1029/2021JB021925 Wei, C., Ye, L., Hu, Y. S., et al., 2019. Distribution and Occurrence of Ge and Related Trace Elements in Sphalerite from the Lehong Carbonate-Hosted Zn-Pb Deposit, Northeastern Yunnan, China: Insights from SEM and LA-ICP-MS Studies. Ore Geology Reviews, 115: 103175. https://doi.org/10.1016/j.oregeorev.2019.103175 Wei, C., Ye, L., Hu, Y. S., et al., 2021a. LA-ICP-MS Analyses of Trace Elements in Base Metal Sulfides from Carbonate-Hosted Zn-Pb Deposits, South China: A Case Study of the Maoping Deposit. Ore Geology Reviews, 130: 103945. https://doi.org/10.1016/j.oregeorev.2020.103945 Wei, C., Ye, L., Huang, Z. L., et al., 2021b. In Situ Trace Elements and S Isotope Systematics for Growth Zoning in Sphalerite from MVT Deposits: A Case Study of Nayongzhi, South China. Mineralogical Magazine, 85(3): 364-378. https://doi.org/10.1180/mgm.2021.29 Wei, Y. K., Li, Z. H., Li, Z. X., et al., 2023. Weather Factors and Prediction Model of Synthetic Electric Field of Ultra-High Voltage Direct-Current Transmission Line. Science Technology and Engineering, 23(9): 3769-3778 (in Chinese with English abstract). Xin, F. X., Whittingham, M. S., 2020. Challenges and Development of Tin-Based Anode with High Volumetric Capacity for Li-Ion Batteries. Electrochemical Energy Reviews, 3(4): 643-655. https://doi.org/10.1007/s41918-020-00082-3 Xu, J., Cook, N. J., Ciobanu, C. L., et al., 2021. Indium Distribution in Sphalerite from Sulfide-Oxide-Silicate Skarn Assemblages: A Case Study of the Dulong Zn-Sn-In Deposit, Southwest China. Mineralium Deposita, 56(2): 307-324. https://doi.org/10.1007/s00126-020-00972-y Yang, R. L., 2018. Analysis of Current Situation of Lead and Zinc Mine Resources Development and Suggestions for Sustainable Development in China. World Nonferrous Metals, (1): 148, 150 (in Chinese with English abstract). Yang, Y., Zhang, H. S., Yang, X. Y., et al., 2024. Evolution of the Hydrothermal Ore-Forming System of Ashele VMS-Type Cu-Zn Deposit in Xinjiang, NW China: Insights from Mineralogy and Geochemistry of Sulfides. Ore Geology Reviews, 167: 105977. https://doi.org/10.1016/j.oregeorev.2024.105977 Zha, W. T., Yan, L. C., Chen, B., et al., 2022. Regional Ultra-Short-Term Wind Power Forecasting Method Based on TPE-LSTM. Computer Applications and Software, 39(11): 25-30, 111 (in Chinese with English abstract). Zhang, C. Q., Liu, H., Wang, D. H., et al., 2015. A Preliminary Review on the Metallogeny of Pb-Zn Deposits in China. Acta Geologica Sinica, 89(4): 1333-1358. https://doi.org/10.1111/1755-6724.12532 Zhang, Y., Han, R. S., Ding, X., et al., 2019. An Experimental Study on Metal Precipitation Driven by Fluid Mixing: Implications for Genesis of Carbonate-Hosted Lead-Zinc Ore Deposits. Acta Geochimica, 38(2): 202-215. https://doi.org/10.1007/s11631-019-00314-4 Zheng, Y., Yu, P. P., Li, Z. K., et al., 2023. Critical Metals Ga, Ge and In in the Global Pb-Zn Deposits: Current Understanding, Challenges and Perspectives. Journal of Earth Science, 34(4): 1308-1311. https://doi.org/10.1007/s12583-023-1909-0 陈钧渝, 沈鸿杰, 颜伟裕, 2023. 中天山狼牙泉铅锌矿床闪锌矿LA-ICP-MS微量元素特征对矿床成因的指示意义. 矿产勘查, 14(3): 377-391. 董赛娜, 王达, 马国桃, 等, 2024. 基于机器学习的闪锌矿微量元素特征在铅锌矿床类型识别中的应用. 成都理工大学学报(自然科学版), 51(4): 614-629. 侯霖莉, 吴松, 易建洲, 等, 2024. 基于机器学习的绿泥石微量元素判别矿床类型. 地球科学, 49(12): 4303-4317. doi: 10.3799/dqkx.2023.173 黄鑫怀, 李增华, 邓腾, 等, 2023. 基于机器学习的华南诸广山花岗岩体铀矿潜力评价. 地球科学, 48(12): 4427-4440. doi: 10.3799/dqkx.2022.006 金露英, 秦克章, 李光明, 等, 2015. 大兴安岭北段岔路口斑岩Mo-热液脉状Zn-Pb成矿系统硫化物微量元素的分布、起源及其勘探指示. 岩石学报, 31(8): 2417-2434. 任涛, 周家喜, 王蝶, 等, 2019. 滇东北富乐铅锌矿床微量元素和S-Pb同位素地球化学研究. 岩石学报, 35(11): 3493-3505. 魏寅孔, 李振华, 李振兴, 等, 2023. 特高压直流输电线路合成电场的天气影响因素及预测模型. 科学技术与工程, 23(9): 3769-3778. 杨荣林, 2018. 浅析我国铅锌矿资源开发现状及可持续发展建议. 世界有色金属, (1): 148, 150. 查雯婷, 闫利成, 陈波, 等, 2022. 基于TPE-LSTM的区域超短期风电功率预测. 计算机应用与软件, 39(11): 25-30, 111. -
陈忠元 附表.xlsx
-
点击查看大图
图(11) / 表(1)
计量
- 文章访问数: 137
- HTML全文浏览量: 17
- PDF下载量: 7
- 被引次数: 0
