-
绝缘强度（Er）是筛选和评估六氟化硫（SF6）替代气体的关键指标。本研究基于机器学习方法，构建了SF6替代气体的Er预测模型。首先，收集了88个气体分子相对SF6的Er数据并计算了这些分子的全局参数和静电势参数，并将其作为描述符。采用5种经过五折交叉验证和超参数优化后的机器学习方法，在Er数据及描述符之间建立Er预测模型。研究表明，自适应增强回归模型表现突出，其决定系数达到了0.90，平均绝对误差和均方根误差分别为0.17和0.18。结合Shapley加性解释量化了描述符特征对Er的贡献，发现极化率是影响Er的主要因素。最后，利用SF6及已知的六种环保替代气体对自适应增强回归模型的Er进行了验证，其绝对误差均在0.02-0.33之间，进一步证实了该模型的可靠性。本研究有望为筛选SF6替代气体提供一种可行的路径。Dielectric strength (Er) is a critical factor in screening and evaluating for SF6 replacement gas. The conventional experimental methods for measuring Er are not only exceptionally time-consuming but also prohibitively expensive. This work constructed an Er prediction model for SF6 replacement gases based on machine learning methods. First, an exhaustive literature survey is performed to collect 88 high-quality experimental Er values. Second, a total of 32 insightful microscopic descriptors are accurately calculated for each compound based on density functional theory, including both global parameters and molecular electrostatic potential parameters. Furthermore, five state-of-the-art machine learning algorithms, which have been carefully modified based on five-fold cross-validation and hyperparameter optimization, are utilized to train and test the 88 experimental Er data and their relevant microscopic descriptors. Finally, the result reveals that Ada Boost regression model demonstrates superior predictive performance with a coefficient of determination of 0.90, a mean absolute error of 0.17, and a root mean square error of 0.18. Moreover, Shapley Additive exPlanations analysis is used to reveal the correlation between the microscopic descriptors and Er. The results indicate that polarizability emerges as the predominant factor significantly affecting Er, which accounts for as high as 17.3%, followed by the molecular weight (14.1%). Specifically, molecules with high α are more prone to deformation under the action of an electric field, and their electron clouds are more likely to be polarized, which has a positive correlation with Er. There is an approximately positive correlation between the molecular weight and the Er of gases. To confirm the reliability of Ada Boost regression model for Er prediction, the Er of SF6 and six known environmentally friendly replacement gases were tested within an absolute error of 0.02-0.33. This study provides a feasible pathway to accelerate the search for SF6 replacement gases.
-
Keywords:
- Dielectric strength prediction /
- Machine learning /
- SF6 replacement gas /
- Density Functional Theory
-
[1] Pan B, Wang G, Shi H, Shen J, Ji H K, Kil G S 2020 Appl. Sci. 10 2526.
[2] Cui Z, Yi L, Song X, Tian S S, Tang J, Hao Y P 2024 Sci. Total Environ 906 167347.
[3] Gao K L, Yang Y, Zhou W J, Wang B S, Ding W D, Lin X, Yan X L, Zhang B Y, Wang D B, Xiao S 2024 CSEE J. Power energy 44 7395 （in Chinese） [高克利，杨圆，周文俊，王宝山，丁卫东，张晓星，林莘，颜湘莲，张博雅，王邸博，肖淞 2024 中国电机工程学报 44 7395]
[4] Ma T, Wang Q G, Wang W M 2025 Low Temperature and Specialty Gases 43 1. （in Chinese） [马腾，王泉高，王伟民 2025 低温与特气 43 1]
[5] Liu W, Dong W C, Song Y M, Zhao Y H, Wu P P, Huang X, Wang K, Cheng L J 2023 Int. J. Quantum Chem. 123, e27114.
[6] Zhang X X, Li Y, Xiao S, Tang J. 2017 Environ. Sci. Technol. 51, 10127.
[7] Li Y, Pang Z Y, Zheng H B 2024 IEEE Trans. Dielectr. Electr. Insul. 31 297.
[8] Chen K, Lin X, Xu J Y, Xiong W, Song H F 2024 High Volt. 60 99. （in Chinese） [陈凯，林莘，徐建源，熊玮，宋会发 2024 高压电器 60 99]
[9] Long Y X, Guo L P, Shen Z Y 2019 High Voltage 45 1064 （in Chinese） [龙云翔，郭立平，沈震宇，陈成，周文俊，刘伟 2019 高电压技术 45 1064]
[10] Liu Y J, Ren M, Wang K 2025 High Voltage 51 1206 （in Chinese） [柳玉洁，任明，王凯，陈旭，王腾腾，董明 2025 高电压技术 51 1206]
[11] Yang Z Q, Zeng J J, Ma Y D, Wei T, Zhao B, Liu Y Z, Zhang W, Lv J, Li X W, Zhang B Y, Tang N, Li L, Sun D W 2023 Chem. Ind. Eng. Prog. 42 4093 （in Chinese） [杨志强，曾纪珺，马义丁，尉涛，赵波，刘英哲，张伟，吕剑，李兴文，张博雅，唐念，李丽，孙东伟 2023 化工进展 42 4093]
[12] Yu X J, Hou H, Wang B S 2017 J. Comput. Chem. 38 721
[13] Brand K P 1982 IEEE Trans. Electr. Insul. 17 451
[14] Zhang B Y, Chen L, Li X W, Guo Z, Pu Y J, Tang N 2020 IEEE Trans. Dielectr. Electr. Insul. 27 1187
[15] Meurice N, Sandre E, Aslanides A, Vercauteren D P 2004 IEEE Trans. Dielectr. Electr. Insul. 11 946
[16] Zhang C H, Shi H X, Cheng L, Zhao K, Xie X Y, Ma H B 2016 IEEE Trans. Dielectr. Electr. Insul. 23 2572.
[17] Heng P P, Zhang M, Hou H, Wang B S 2024 Chem. J. Chinese U. 45 61 （in Chinese） [衡盼盼，张咪，侯华，王宝山 2024 高等学校化学学报 45 61]
[18] Liu G P, Yang S, Zhang N N, Wang H, Xiao J X 2022 High Voltage 48 2208 （in Chinese） [刘关平，杨帅，张闹闹，王航，肖集雄 2022 高电压技术 48 2208]
[19] Zhou W J, Qiu R, Gao K L, Zheng Y, Hou H, Wang B S, Luo Y B, Yu J H 2023 High Voltage 49 895 （in Chinese） [周文俊，邱睿，高克利，郑宇，侯华，王宝山，罗运柏，喻剑辉 2023 高电压技术 49 895]
[20] Zhao Q L, Tong D J, Tu Y P, Zheng Z, Qiu X Y, Wang C 2024 IEEE Trans. Dielectr. Electr. Insul. 30 49
[21] Li H M, Zhao S, Xiao D M 2024 IEEE Trans. Dielectr. Electr. Insul. 31 2013
[22] Zhang M, Hou H, Wang B S 2023 High Voltage 9 484-494.
[23] Rabie M, Dahl D A, Donald S M A, Reiher M, Franck C M 2013 IEEE Trans. Dielectr. Electr. Insul. 20 856.
[24] Sun H, Liang L Q, Wang C L, Wu L, Yang F, Rong M Z 2020 IEEE Access 8 124204
[25] Zhao G B, Kim H W, Yang C W, Chung Y G 2024 J. Phys. Chem. A 128 2399
[26] Liu W, Zha J W, Ling M X, Li D, Shen K D, Cheng L J 2024 Chem. Phys. 588 112447.
[27] Štrumbelj E, Kononenko I 2014 Knowledge and Information Systems 41 647
[28] Zhang Q, Zheng Y J, Sun W B, Ou Z P, Odunmbaku O, Li M, Chen S S, Zhou Y L, Li J, Qin B, Sun K 2022 Adv. Sci. 9 6 2104742.
[29] Luo H R, Gou Q Z, Zheng Y J, Wang K X, Yuan R D, Zhang S D, Fang W, Luogu Z, Hu Y Z, Mei H, Song B, Sun K, Wang J. Li M 2025 ACS Nano 19 2 2427-2443.
[30] Zhang Q, Tan W, Ning Y Q, Nie G Z, Cai M Q, Wang J N, Zhu H P, Zhao Y Q 2024 Acta Phys. Sin. 73 7 （in Chinese） [张桥，谭薇，宁勇祺，聂国政，蔡孟秋，王俊年，朱慧平，赵宇清 2024 物理学报, ,73 7]
[31] Li X W, Zhao H, Murphy A 2018 J. Phys. D. Appl. Phys. 51 153001
[32] Meurice N, Sandre E, Aslanides A, Vercauteren D P 2004 IEEE Trans. Dielectr. Electr. Insul. 11 946
[33] Yang S, Wu S B, Chen H 2024 Comput. Theor. Chem. 1238 114680
[34] Wu S B, Yang S, Chen H 2024 J. Mol. Model 30 413
[35] Lu T, Chen F. W. 2012 J. Mol. Graph. Model. 38, 314-323
[36] Lu T, Chen F J 2012 Comput. Chem. 33 580
[37] Humphrey W, Dalke A, Schulten K 1996 J. Mol. Graph. 14 33
[38] Gillet A, Cher S, Tassé M, Blon T, Alves S, Izzet G, Chaudret B, Proust. A, Demont P, Volatron F, Tricard S 2021 Nanoscale Horiz. 6 271
[39] Vijh A K 1982 IEEE Trans. Dielectr. Electr. Insul. EI-17 84
计量
- 文章访问数: 21
- PDF下载量: 0
- 被引次数: 0
下载:
