Mô Hình Dự Đoán Độ Ẩm Than Dựa Trên Machine Learning Với Mẫu Nhỏ

Jingyu Wang1,2,3, Shuheng Tang2,3,1, Songhang Zhang1,2,3, Zhaodong Xi1,2,3, Jianwei Lv4,3
1Coal Reservoir Laboratory of National Engineering Research Center of CBM Development and Utilization, China University of Geosciences, Beijing, China
2Key Laboratory of Marine Reservoir Evolution and Hydrocarbon Enrichment Mechanism, Ministry of Education, China University of Geosciences, Beijing, China
3Institute of Energy Resources, China University of Geosciences, Beijing, China
4Chinese Academy of Natural Resources Economics, Beijing, China

Tóm tắt

Trong lĩnh vực kiểm soát bụi than và phát triển khí methane từ tầng than (CBM), độ ẩm là một tham số quan trọng của than, thường được xác định bởi góc tiếp xúc giữa than và nước (CA). Để xây dựng một mô hình dự đoán CA chính xác, dữ liệu lớn về các thành phần công nghiệp, hàm lượng nguyên tố và CA của than đã được thu thập. Hai tập dữ liệu đã được sử dụng: một tập lớn gồm 98 nhóm dữ liệu được thu thập từ nhiều nguồn khác nhau, và một tập nhỏ với 16 nhóm dữ liệu được thu thập từ một nguồn duy nhất. Những tập dữ liệu này đã được sử dụng để phát triển các mô hình bằng ba phương pháp học máy (ML): K-láng giềng gần nhất, hồi quy vector hỗ trợ, và mạng nơ-ron hồi tiếp. Kết quả cho thấy hiện tượng giảm độ chính xác đáng kể trong cả ba phương pháp ML khi áp dụng cho tập dữ liệu huấn luyện và thử nghiệm của mẫu lớn. Hiện tượng giảm này chủ yếu do sự khác biệt trong cách xử lý mẫu than giữa các nhà nghiên cứu khác nhau. Ngược lại, cả ba phương pháp ML đều thể hiện sự quá khớp rõ rệt trên tập dữ liệu huấn luyện của mẫu nhỏ, dẫn đến khả năng tổng quát hạn chế trên tập dữ liệu thử nghiệm. Hạn chế này phát sinh từ lượng dữ liệu ít ỏi trong mẫu nhỏ. Để giải quyết vấn đề này, kỹ thuật tăng cường mẫu thiểu số đã được sử dụng để tạo ra các mẫu mở rộng cho mẫu nhỏ. Mối tương quan xác định của các mẫu mở rộng dao động từ 0.82 đến 0.92, cho thấy độ khớp rất tốt. Thêm vào đó, tỷ lệ ưu việt của độ khớp của các tập dữ liệu huấn luyện và thử nghiệm nằm trong khoảng từ 0.92 đến 1. Phương pháp này hiệu quả trong việc tránh rủi ro giảm độ chính xác trong các tập dữ liệu mẫu lớn và quá khớp trong các tập dữ liệu huấn luyện mẫu nhỏ. Ở giai đoạn cuối cùng, mô hình đã phát triển được sử dụng để dự đoán độ ẩm của mẫu than từ ba mỏ than trong lưu vực Qinshui, Trung Quốc. Giá trị CA dự đoán cho thấy sự nhất quán cao với các giá trị CA được đo trong phòng thí nghiệm. Nghiên cứu toàn diện này đã phân tích kỹ lưỡng các lý do cơ bản gây ra sự thất bại của các mô hình ML trong việc xử lý hiệu quả dữ liệu mẫu lớn và nhỏ đối với CA. Nó cũng cung cấp một giải pháp quý giá cho những vấn đề nêu trên bằng cách tăng cường dữ liệu với các mẫu nhỏ, điều này có ý nghĩa lớn trong việc cho phép dự đoán nhanh chóng và chính xác các giá trị CA bằng cách sử dụng dữ liệu tham số than hạn chế.

Từ khóa

#độ ẩm than #mô hình dự đoán #học máy #góc tiếp xúc #tăng cường dữ liệu

Tài liệu tham khảo

Altman, N. S. (1992). An introduction to kernel and nearest-neighbor nonparametric regression. The American Statistician, 46(3), 175–185. Arif, M., Barifcani, A., Lebedev, M., & Iglauer, S. (2016). CO2-wettability of low to high rank coal seams: Implications for carbon sequestration and enhanced methane recovery. Fuel, 181, 680–689. Cao, W., Qin, Q., Cao, W., Lan, Y., Chen, T., Xu, S., & Cao, X. (2017). Experimental and numerical studies on the explosion severities of coal dust/air mixtures in a 20-L spherical vessel. Powder Technology, 310, 17–23. Chawla, N. V., Bowyer, K. W., Hall, L. O., & Kegelmeyer., W. P. (2002). SMOTE: Synthetic Minority Over-sampling Technique. Journal of Artificial Intelligence Research, 16(1), 321–357. Chen, X., Gao, J., Deng, C., Ge, S., Fan, C., & Zhang, W. (2022). Experimental study on chemical structure and wetting influence of imidazole ionic liquids on coal. Fuel, 330, 125545. Chen, Y., Ma, Z., Ma, D., Li, W., Li, G., Yang, F., Zheng, C., Teng, J., & Ji, Y. (2021). Effects of wettability differences of different macroscopic composition of coal on methane adsorption and desorption. Coal Science and Technology, 49(11), 47–55. Chen, Y., Xia, W., & Xie, G. (2018). Contact angle and induction time of air bubble on flat coal surface of different roughness. Fuel, 222, 35–41. Cover, T. (1968). Estimation by the nearest neighbor rule. IEEE Transactions on Information Theory, 14(1), 50–55. Durucan, S., Ahsan, M., Syed, A., Shi, J.-Q., & Korre, A. (2013). Two phase relative permeability of gas and water in coal for enhanced coalbed methane recovery and CO2 storage. Energy Procedia, 37, 6730–6737. Glanville, J. O., & Wightman, J. P. (1979). Actions of wetting agents on coal dust. Fuel, 58(11), 819–822. Hanning, R. N., & Rutter, P. R. (1989). A simple method of determining contact angles on particles and their relevance to flotation. International Journal of Mineral Processing, 27(1), 133–146. Hao, J. (2016). Effect of surfactants on the wettability of coal [Master's degree, AnHui University of Science and Technology]. Hao, P. (2019). Influence of High Pressure with Fracturing Fluid on Microstructure of Low-medium Rank Coal Reservoir [Master's degree, Taiyuan University of Technology]. Hu, Q. (2020). Experimental Study on Effect of Surface Active Ionic Liquids on Coal Dust Wettability [Master's degree, China University of Mining and Technology]. Hu, H., & Sun, Y. (2016). Effect of nanostructures on heat transfer coefficient of an evaporating meniscus. International Journal of Heat and Mass Transfer, 101, 878–885. Ibrahim, A. F. (2022). Application of various machine learning techniques in predicting coal wettability for CO2 sequestration purpose. International Journal of Coal Geology, 252, 103951. Ibrahim, A. F. (2023). Prediction of shale wettability using different machine learning techniques for the application of CO2 sequestration. International Journal of Coal Geology, 276, 104318. Ibrahim, A. F., & Nasr-El-Din, H. A. (2016). Effect of water salinity on coal wettability during CO2 sequestration in coal seams. Energy & Fuels, 30(9), 7532–7542. Jia, L., Li, K., Shi, X., Zhao, L., & Linghu, J. (2021). Application of gas wettability alteration to improve methane drainage performance: A case study. International Journal of Mining Science and Technology, 31(4), 621–629. Jia, L., Li, K., Zhou, J., Yan, Z., Wang, Y., & Mahlalela, B. M. (2019). Experimental study on enhancing coal-bed methane production by wettability alteration to gas wetness. Fuel, 255, 115860. Jiang, B., Yu, C.-F., Yuan, L., Lu, K., Tao, W., Lin, H., & Zhou, Y. (2023). Investigation on oxidative pyrolysis characteristics of bituminous coal through thermal analysis and density functional theory. Applied Energy, 349, 121680. Jiang, B., Zhou, Y., Ji, B., Zheng, Y., Yu, C.-F., Huang, J., & Wang, X.-H. (2023b). Investigation on the effect of functional groups on the wettability of coal dust: Experiments and theoretical validation. Fuel, 351, 128987. Jin, H., Zhang, Y., Dong, H., Zhang, Y., Sun, Y., Shi, J., & Li, R. (2022). Molecular dynamics simulations and experimental study of the effects of an ionic surfactant on the wettability of low-rank coal. Fuel, 320, 123951. Kong, D., Ma, L., Li, Y., & Liu, C. (2019). Effect of physical properties and chemical composition of low rank coal on its wettability. Coal Processing & Comprehensive Utilization, 238(05), 36–41+38. Leelamanie, D. A. L., Karube, J., & Yoshida, A. (2008). Characterizing water repellency indices: Contact angle and water drop penetration time of hydrophobized sand. Soil Science and Plant Nutrition, 54(2), 179–187. Li, J. Y. (2016). The influence factors of coal surface wettability [Master degree thesis, Henan Polytechnic University]. Li, C., Zhang, J., Han, J., & Yao, B. (2021). A numerical solution to the effects of surface roughness on water-coal contact angle. Scientific Reports, 11(1), 459. Li, J., & Li, K. (2016). Influence factors of coal surface wettability. Journal of China Coal Society, 41(S2), 448–453. Li, P., Ma, D., Zhang, H., Li, W., & Yang, F. (2016). Influence of high and low rank coal wettability and methane adsorption/desorption characteristics. Coal Geology & Exploration, 44(5), 80–85. Liao, X., Wang, B., Wang, L., Zhu, J., Chu, P., Zhu, Z., & Zheng, S. (2021a). Experimental study on the wettability of coal with different metamorphism treated by surfactants for coal dust control. ACS Omega, 6(34), 21925–21938. Liao, X., Wang, L., Zhu, J., Chu, P., Liu, Q., & Yang, T. (2021b). Experimental study on the wettability of tectonic soft coal in Huaibei mining area China. Energy & Fuels, 35(8), 6585–6599. Liu, S., Yang, Y., Yu, L., Zhu, F., Cao, Y., Liu, X., & Cao, Y. (2022). Predicting gas production by supercritical water gasification of coal using machine learning. Fuel, 329, 125478. Liu, Z., Zhou, G., Li, S., Wang, C., Liu, R., & Jiang, W. (2020). Molecular dynamics simulation and experimental characterization of anionic surfactant: Influence on wettability of low-rank coal. Fuel, 279, 118323. Lu, Y., Liu, D., Cai, Y., Gao, C., Jia, Q., & Zhou, Y. (2022). AFM measurement of roughness, adhesive force and wettability in various rank coal samples from Qinshui and Junggar basin China. Fuel, 317, 123556. Lu, Y., Liu, D., Cai, Y., & Zhou, Y. (2023). Novel discrete fracture network model for multiphase flow in coal. Energy & Fuels, 37(3), 1907–1916. Ma, Y. (2016). Experimental study on wettability of coal dust and new wetting agent [Master's degree, AnHui University of Science and Technology]. McCollum, D. L. (2021). Machine learning for energy projections. Nature Energy, 6(2), 121–122. Niu, C., Xia, W., & Peng, Y. (2018). Analysis of coal wettability by inverse gas chromatography and its guidance for coal flotation. Fuel, 228, 290–296. Qi, J. (2017). Study on Wettability and Floatability of Different Rank Coal [Master's degree, Taiyuan University of Technology]. Shen, J., Qin, Y., Li, Y., & Wang, G. (2019). Experimental investigation into the relative permeability of gas and water in low-rank coal. Journal of Petroleum Science and Engineering, 175, 303–316. Shi, B., & Dhir, V. K. (2009). Molecular dynamics simulation of the contact angle of liquids on solid surfaces. The Journal of Chemical Physics, 130(3), 034705. Shi, Q., Qin, B., Bi, Q., & Qu, B. (2018). Changes in the surface structure of coal caused by igneous intrusions and their effect on the wettability. Energy & Fuels, 32(9), 9371–9379. Sun, Y. (2014). Study on wettability of coal dust and complex program of dust agent [Master's degree, AnHui University of Science and Technology]. Sun, X., Yao, Y., Chen, J., Xie, S., & Li, C. (2015). Determination of coal wettability by using low-field nuclear magnetic resonance. Geoscience, 29(1), 190–197. Sun, X. X., Yao, Y. B., Liu, D. M., & Zhou, Y. F. (2018). Investigations of CO2-water wettability of coal: NMR relaxation method. International Journal of Coal Geology, 188, 38–50. Tan, X. (2019). Experimental study on wetting characteristics of coal dust and efficiency of spray dustfall [Master's degree, Hunan University of Science and Technology]. Vo Thanh, H., Rahimi, M., Dai, Z., Zhang, H., & Zhang, T. (2023). Predicting the wettability rocks/minerals-brine-hydrogen system for hydrogen storage: Re-evaluation approach by multi-machine learning scheme. Fuel, 345, 128183. Wang, Z. (2008). Study of The Coal Dust Wettability and Adsorption Characteristics [Master's degree, Shandong University of Science and Technology]. Wang, X. (2020). Study on synergistic effect of surfactant compounding on wettability of coal dust [Doctor's degree, AnHui University of Science and Technology]. Wang, Y. (2020). Quantitative Research on the wetting effect of surfactant oncoal dust [Master's degree, Xi`an University of Science and Technology]. Wang, K., Ding, C., Jiang, S., Zhengyan, W., Shao, H., & Zhang, W. (2019a). Application of the addition of ionic liquids using a complex wetting agent to enhance dust control efficiency during coal mining. Process Safety and Environmental Protection, 122, 13–22. Wang, L., Liao, X., Zha, M., Zhang, R., Pei, X., & Pan, R. (2020). Study on wetting characteristics of coal dust in soft coal based on principal component analysis. Coal Science and Technology, 48(2), 104–109. Wang, P., Tan, X., Zhang, L., Li, Y., & Liu, R. (2019b). Influence of particle diameter on the wettability of coal dust and the dust suppression efficiency via spraying. Process Safety and Environmental Protection, 132, 189–199. Wang, Y., Kiziltas, A., Blanchard, P., & Walsh, T. R. (2022). ContactAngleCalculator: An automated, parametrized, and flexible code for contact angle estimation in visual molecular dynamics. Journal of Chemical Information and Modeling, 62(24), 6302–6308. Wei, D., Chander, S., & Hogg, R. (1992). Distribution of wettability of coal. Coal Preparation, 10(1–4), 37–45. Xi, X., Jiang, S., Zhang, W., Wang, K., Shao, H., & Wu, Z. (2019). An experimental study on the effect of ionic liquids on the structure and wetting characteristics of coal. Fuel, 244, 176–183. Yan, J. (2019). Study on wettability of coal dust based on its surface properties [Doctor's degree, China University of Mining & Technology, Beijing]. Yan, M., Luo, H., Yang, T., Yan, D., Wei, J., Lin, H., & Li, S. (2023). Experimental study on fractal characteristics of surface roughness of briquettes and their effect on wettability of coal samples. Natural Resources Research, 32(3), 1235–1249. Yan, M., Yue, M., Lin, H., Yan, D., Wei, J., Qin, X., & Zhang, J. (2021). Experimental study on the influence of middle and low rank coal functional groups on coal wettability. Coal Science and Technology. https://doi.org/10.13199/j.cnki.cst.2021-1097 Yang, J. (2008). Study on Wettability Mechanism of Coal Dust [Doctor's degree, Shandong University of Science and Technology]. Ye, Y., Tang, S., Xi, Z., Jiang, D., & Duan, Y. (2022). A new method to predict brittleness index for shale gas reservoirs: Insights from well logging data. Journal of Petroleum Science and Engineering, 208, 109431. Yuan, M., Nie, W., Yu, H., Yan, J., Bao, Q., Zhou, W., Hua, Y., Guo, L., & Niu, W. (2021). Experimental and molecular dynamics simulation study of the effect of different surfactants on the wettability of low-rank coal. Journal of Environmental Chemical Engineering, 9(5), 105986. Zhang, K. (2017). Study on the wettability of No.4 coal in Dafosi coal field [Master's degree, Xi’an University of Science and Technology]. Zhang, R. (2019). Study on Coal Dust Wetting Characteristic and Dust-proof Technology in Driving Face of Outburst Coal Seam [Master's degree, China University of Mining and Technology]. Zhang, K. (2020). Study of Wetting Characteristics of Coal Body Active Agent Based on Surface Roughness [Master's degree, Henan Polytechnic University]. Zhang, Y. (2021). Characteristics of coal reservoir retention effect and its influencing factors [Master's degree, Taiyuan University of Technology]. Zhang, H., Ma, D., & Liu, H. (2019). Comparative study on wettability of 4# coal from Dafosi and Hujiahe Coalfield. Journal of Xi’an University of Science and Technology, 39(03), 435–442. Zhang, J., Li, H., Liu, Y., Li, X., Xie, J., Dai, Z., Ye, S., Li, L., Zhou, W., Zhao, Y., & Hao, H. (2021a). Micro-wetting characteristics of coal dust and preliminary study on the development of dust suppressant in Pingdingshan mining area. Journal of China Coal Society, 46(3), 812–825. Zhang, K., Yan, M., Li, S., Wei, J., & Zhou, M. (2021b). Experimental study on the influence of coal surface roughness on wettability of non-cationic surfactants. Safety in Coal Mines, 52(11), 8–15. Zhang, R., Xing, Y., Xia, Y., Luo, J., Tan, J., Rong, G., & Gui, X. (2020). New insight into surface wetting of coal with varying coalification degree: An experimental and molecular dynamics simulation study. Applied Surface Science, 511, 145610. Zhang, X., Xu, C., Yan, G., & Huang, Q. (2015). Influencing factors analysis of different types of coal wettability. Safety in Coal Mines, 46(01), 156–158+162. Zhang, Y., Li, Q., Hu, Q., Zhai, C., Song, M., Xu, J., Deng, Y., Liu, P., Sun, Y., Shi, J., & Hu, L. (2022). Pore wetting process characterization of equal-sized granular coals by using LF-NMR technology. Fuel, 313, 122670. Zhao, D. (2014). Study on Wettability of coal and Its Matching Performance with Wetting Dust-fall Agent [Master's degree, China University of Mining and Technology]. Zhao, H. (2021). Effect of Clean Fracturing Fluid on Adsorption and Desorption of Medium and High Rank Coal [Master's degree, China University of Mining and Technology]. Zhou, G., Qiu, H., Zhang, Q., Xu, M., Wang, J., & Wang, G. (2016). Experimental investigation of coal dust wettability based on surface contact angle. Journal of Chemistry, 2016, 9452303. Zhu, C., Li, G., Xing, Y., & Gui, X. (2021). Adhesion forces for water/oil droplet and bubble on coking coal surfaces with different roughness. International Journal of Mining Science and Technology, 31(4), 681–687. Zhu, Q. (2022). Theoretical basis of dredging and efficient development of high−rank coalbed methane in China: A case study of the Qinshui Basin. Coal Geology & Exploration, 50(03), 82–91. Zosky, G. R., Hoy, R. F., Silverstone, E. J., Brims, F. J., Miles, S., Johnson, A. R., Gibson, P. G., & Yates, D. H. (2016). Coal workers’ pneumoconiosis: an Australian perspective. Medical Journal of Australia, 204(11), 414–418.