机器学习预测Cs在缓冲材料和岩石上的分配系数
Prediction of Cs Distribution Coefficients on Buffer Materials and Rocks by Machine Learning Methods
DOI: 10.12677/NST.2023.113022, PDF, 下载: 287  浏览: 1,164  国家自然科学基金支持
作者: 何文科, 陈 涛*：华北电力大学核科学与工程学院，北京;华北电力大学非能动核能安全技术北京市重点实验室，北京
关键词: 机器学习；吸附；Cs；分配系数；安全评价；Machine Learning； Adsorption； Cs； Distribution Coefficient； Safety Assessment
摘要: 缓冲材料和天然岩石是高放废物深层地质处置库多重屏障系统的重要组成部分。核素在缓冲材料和围岩中的分配系数(Kd)是处置库安全评价中的重要参数之一。本研究采用4种常见的基于树的机器学习集成算法预测Cs在缓冲材料和岩石上的Kd。结果表明，RF模型的预测性能最好，其中对准确预测Kd最敏感的因素是t。LSR、CEC和T对Kd的影响都很小。本工作证明了机器学习在促进高放废物处置库安全评价方面的应用前景。
Abstract: Buffer materials and natural rocks are important components of the multiple barriers system of the deep geological repository for high-level radioactive waste (HLW). Distribution coefficient (Kd) of nuclides in buffer materials and surrounding rocks plays an important role in the performance and safety assessment of HLW repository. In this study, four common tree-based machine learning integration algorithms were used to predict the Kd of Cs on buffer materials and rocks. The results indicate that RF model has the best predictive performance, and t is the most sensitive factor to accurately predict Kd. The effects of LSR, CEC and T on Kd were small. This work demonstrates the great application prospect of machine learning in the performance and safety assessment of HLW repository.
文章引用：何文科, 陈涛. 机器学习预测Cs在缓冲材料和岩石上的分配系数[J]. 核科学与技术, 2023, 11(3): 210-217. https://doi.org/10.12677/NST.2023.113022

参考文献

[1] 彭峥, 韩阳, 刘竞戈. 高放射性废物地下实验室施工风险管理模式[J]. 四川建筑, 2016, 36(4): 258-260.
[2] 秦爱芳, 贾旭. 初始孔隙率对高放射性废物处置库缓冲材料性状的影响[J]. 上海大学学报(自然科学版), 2019, 25(4): 516-525.
[3] 王安, 艾英钵, 曹炜炜, 范志浩. Pb2+在高岭土土柱中迁移时吸附特性的研究[J]. 粉煤灰综合利用, 2022, 36(1): 32-38.
[4] Stockmann, M., Schikora, J., Becker, A., et al. (2017) Smart Kd-Values, Their Uncer-tainties and Sensitivities—Applying a New Approach for Realistic Distribution Coefficients in Geochemical Modeling of Complex Systems. Chemosphere, 187, 277-285.
https://doi.org/10.1016/j.chemosphere.2017.08.115
[5] Britz, S., Noseck, U., Durner, W., et al. (2016) Eu(III) Reactive Transport Modeling: The “Smart Kd” Approach. Goldschmidt Conference, Yokohama, 27 June 2016, 307.
[6] Chen, M., Hao, Y., Hwang, K., Wang, L. and Wang, L. (2017) Dis-ease Prediction by Machine Learning over Big Data from Healthcare Communities. IEEE Access, 5, 8869-8879.
https://doi.org/10.1109/ACCESS.2017.2694446
[7] Marturana, F. and Fabio, S. (2013) A Machine Learn-ing-Based Triage Methodology for Automated Categorization of Digital Media. Digital Investigation, 10, 193-204.
https://doi.org/10.1016/j.diin.2013.01.001
[8] Shariatmadari, N., Farrokhi, M. and Falamaki, A. (2004) Prediction of Partition Coefficients for Cadmium by Neural Networks. 57th Canadian Geotechnical Conference, Quebec City, 1 July 2004, 36-41.
[9] Falamaki, A. (2013) Artificial Neural Network Application for Predicting Soil Distribution Coefficient of Nickel. Journal of Environmental Radioactivity, 115, 6-12.
https://doi.org/10.1016/j.jenvrad.2012.06.008
[10] Da, T., Ren, H., He, W., Gong, S. and Chen, T. (2022) Predic-tion of Uranium Adsorption Capacity on Biochar by Machine Learning Methods. Journal of Environmental Chemical Engineering, 10, Article ID: 108449.
https://doi.org/10.1016/j.jece.2022.108449
[11] Da, T., Chen, T., He, W., et al. (2022) Applying Machine Learning Methods to Estimate the Thermal Conductivity of Bentonite for a High-Level Radioactive Waste Repository. Nuclear En-gineering and Design, 392, Article ID: 111765.
https://doi.org/10.1016/j.nucengdes.2022.111765
[12] Elith, J., Leathwick, J.R. and Hastie, T. (2008) A Working Guide to Boosted Regression Trees. Journal of Animal Ecology, 77, 802-813.
https://doi.org/10.1111/j.1365-2656.2008.01390.x
[13] Sugiura, Y., Suyama, T., Tachi, Y. (2020) Development of JAEA Sorption Database (JAEA-SDB): Update of Sorption/QA Data in FY2019. Japan Atomic Energy Agency, Ibaraki.
[14] 陈廷方, 易发成, 冯启明, 李虎杰. 北川埃洛石黏土对Sr、Co、Cs的吸附性能研究[J]. 中国矿业, 2011, 20(3): 74-77.
[15] Wang, R., Chu, Y. and Chen, M. (2017) Adsorption Kinetics of 137Cs+/90Sr2+ on Ca-Bentonite. Water Environment Research, 89, 791-797.
https://doi.org/10.2175/106143017X14902968254593