搜索

x

留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

基于机器学习的托卡马克偏滤器靶板热负荷预测研究

吴阳海 杜海龙 薛雷 李佳鲜 薛淼 郑国尧

引用本文:
Citation:

基于机器学习的托卡马克偏滤器靶板热负荷预测研究

吴阳海, 杜海龙, 薛雷, 李佳鲜, 薛淼, 郑国尧

Machine learning-based prediction of heat load on Tokamak divertor target plates

WU Yanghai, DU Hailong, XUE Lei, LI Jiaxian, XUE Miao, ZHENG Guoyao
Article Text (iFLYTEK Translation)
PDF
HTML
导出引用
  • 本文首次针对HL-3装置, 采用机器学习方法预测偏滤器靶板等离子体参数, 为未来快速预测大型聚变堆偏滤器热负荷奠定基础. 将机器学习应用于边缘等离子体物理中, 可以显著缩短大型边缘程序SOLPS-ITER模拟所需的时间, 从几周、几个月甚至半年缩短至毫秒级. 研究发现, 通过增加内外偏滤器区的辐射损失作为模型的输入参数, 能够明显提高预测精度(超过90%), 同时增强训练模型的适用性, 可以同时精确预测内外偏滤器靶板热流, 并验证了该特征参数与偏滤器靶板物理量之间的依赖关系. 该工作不仅为偏滤器物理研究提供了有效的方法, 也为未来跨装置预测偏滤器靶板参数提供了坚实的基础.
    The SOLPS-ITER edge plasma simulation code has become a primary tool for designing divertor physics and predicting target plate heat load in fusion research. However, SOLPS-ITER-based divertor design requires not only substantial computational time but also intensive hardware resources, which fundamentally limits its application in advancing divertor optimization, particularly in large-scale fusion reactor divertor design. In this work, the machine learning method is used for the first time to predict the plasma parameters of the divertor target plate for HL-3, which provides a theoretical basis for predicting the heat load of divertor in large fusion reactor in the future. Based on the simulation of the edge plasma code SOLPS-ITER, we first build a database of HL-3 edge plasma parameters, including the upstream inner/outer midplane region and divertor target region. Then, we use the machine learning method and combine with the database to develop an artificial neural network model. Finally, the artificial neural network is used to train a model by using the boundary plasma parameters of the HL-3 device, and the heat load of the divertor target plate is predicted by the given upstream plasma parameters.This work can effectively shorten the time of simulating edge plasma by SOLPS-ITER code from weeks, months or even half a year to several milliseconds. In this work, a multi-layer perceptron (MLP) model is established with different input parameters to predict the electron temperature, density, and parallel heat fluxes of the inner and outer divertor target plates. It is found that reasonably increasing upstream plasma parameters as inputs to the model can not only improve the model’s generalization ability and the accuracy of prediction (both reaching over 90%), but also verify the correlation between upstream plasma parameters and divertor target physical quantities. In addition, a more stable ResMLP model is established on the basis of MLP. This work demonstrates the feasibility of using the neural networks to predict the heat load of the divertor target plate.
  • 图 1  训练和预测的基本流程

    Fig. 1.  Schematic of the basic workflow for training and prediction.

    图 2  机器学习网络结构示意图

    Fig. 2.  Schematic diagram of the machine learning network structure.

    图 3  不同数量的上游参数作为模型输入的损失值图像

    Fig. 3.  Loss values with different numbers of upstream parameters as model inputs.

    图 4  不同壁粒子再循环对应的偏滤器外靶板电子密度和电子温度曲线图

    Fig. 4.  Electron density and temperature profiles at the outer divertor target plate for different wall particle recycling conditions.

    图 5  不同数量的上游参数作为模型输入的测试集相关系数

    Fig. 5.  Test correlation coefficients versus upstream parameter count.

    图 6  外靶板的电子密度、温度和平行热流预测值、真实值

    Fig. 6.  Predicted and true values of electron density, electron temperature, and parallel heat flux on the outer target.

    图 7  残差网络的原理图

    Fig. 7.  Schematic diagram of residual network principles.

    图 8  ResMLP模型的测试集相关系数

    Fig. 8.  Test set correlation coefficients of the ResMLP model.

    表 1  模型的详细输入参数

    Table 1.  Detailed input parameters of the model.

    图3
    图5
    输入参数
    壁粒子再循环系数
    (re)
    外中平面电子密度
    (nomp)
    外中平面电子温度
    (Tomp)
    外中平面平行热流
    (qomp)
    外偏滤器辐射损失
    (raouter)
    内中平面电子密度
    (nimp)
    内中平面电子温度
    (Timp)
    内中平面平行热流
    (qimp)
    内偏滤器辐射损失
    (rainner)
    (a) 4输入
    (b) 6输入
    (c) 8输入
    (d) 5输入
    (e) 7输入
    (f) 9输入
    下载: 导出CSV
  • [1]

    高翔, 万元熙, 丁宁, 彭先觉 2018 中国工程科学 20 25

    Gao X, Wan Y X, Ding N, Peng X J 2018 SSCAE 20 25

    [2]

    Kuang A Q, Ballinger S, Brunner D, Canik J, Creely A J, Gray T, Greenwald M, Hughes J W, Irby J, LaBombard B, Lipschultz B, Lore J D, Reinke M L, Terry J L, Umansky M, Whyte D G, Wukitch S 2020 J. Plasma Phys. 86 865860505Google Scholar

    [3]

    Du H L, Li J X, Xue L, Bonnin X, Zheng G Y, Xiao G L, Fan D M, Tong R H, Xue M, Song X, Wang S, Wu N, Ji X Q, Chen W, Zhong W L 2025 Nucl. Fusion 65 036023Google Scholar

    [4]

    Shi B, Yang Z D , Zhang B, Yang C, Gan K F, Chen M W, Yang J H, Zhang H, Qi LJ L, Gong X Z, Zhang X D, Wang W H 2017 Chin. Phys. Lett. 34 095201

    [5]

    Pitts R A, Kukushkin A, Loarte A, Martin A, Merola M, Kessel C E, Komarov V, Shimada M 2009 Phys. Scr. T 138 014001

    [6]

    Hayashi Y, Masuzaki S, Kobayashi M, Kawamura G, Mukai K, Tanaka H, Murase T 2025 Plasma Phys. Control. Fusion 67 025010Google Scholar

    [7]

    杜海龙, 桑超峰, 王亮, 孙继忠, 刘少承, 汪惠乾, 张凌, 郭后扬, 王德真 2013 物理学报 62 245206Google Scholar

    Du H L, Sang C F, Wang L, Sun J Z, Liu S C, Wang H Q, Zhang L, Guo H Y, Wang D Z 2013 Acta Phys. Sin. 62 245206Google Scholar

    [8]

    Chen W T, Sun J Z, Gao F, Peng L, Wang D Z 2022 Chin. Phys. B 31 075204Google Scholar

    [9]

    Wang M , Nie Q Y, Huang T, Wang X G, Zhang Y J 2024 Chin. Phys. B 33 035204

    [10]

    Si H, Ding R, Senichenkov I, Rozhansky V, Molchanov P, Liu X J, Jia G Z, Sang C F, Mao S F, Chan V 2022 Nucl. Fusion 62 026031Google Scholar

    [11]

    Peng L, Sun Z, Sun J Z, Maingi R, Gao F, Bonnin X, Chang H Y, Wang W K, Liu J Y 2024 Chin. Phys. B 33 115201Google Scholar

    [12]

    Moscheni M, Meineri C, Wigram M, Carati C, De Marchi E, Greenwald M, Innocente P, LaBombard B, Subba F, Wu H, Zanino R 2022 Nucl. Fusion 62 056009

    [13]

    Coster D P 2016 Contrib. Plasma Phys. 56 790Google Scholar

    [14]

    马锐垚, 王鑫, 李树, 勇珩, 上官丹骅 2024 物理学报 73 072802Google Scholar

    Ma R Y, Wang X, Li S, Yong H, Shangguan D H 2024 Acta Phys. Sin. 73 072802Google Scholar

    [15]

    Li H, Fu Y L, Li J Q, Wang Z X 2023 Chin. Phys. Lett 40 125201Google Scholar

    [16]

    陈俊杰, 胡文慧, 肖建元, 郭笔豪, 肖炳甲 2020 计算机系统应用 29 21

    Chen J J, Hu W H, Xiao J Y, Guo B H, Xiao B J 2020 Comput. Syst. Appl. 29 21

    [17]

    Liu B, Wang Y X 2024 Chin. Phys. B 33 084401Google Scholar

    [18]

    Yan Z P, Wang Q Y, Yu X Q, Li J Z 2024 Chin. Phys. Lett. 41 098901Google Scholar

    [19]

    Yue J C, Chen R K, Ma D K, Hu S Q 2025 Chin. Phys. Lett. 42 036301Google Scholar

    [20]

    Bonnin X, Dekeyser W, Pitts R, Coster D, Voskoboynikov S, Wiesen S 2016 Plasma Fusion Res. 11 1403102Google Scholar

    [21]

    Schneider R, Bonnin X, Borrass K, Coster D P, Kastelewicz H, Reiter D, Rozhansky V A, Braams B J 2006 Contrib. Plasma Phys. 46 3Google Scholar

    [22]

    Sytova E, Senichenkov I, Kaveeva E, Rozhansky V, Veselova I, Voskoboynikov S, Coster D 2016 43rd EPS Conference on Plasma Physics Liege, Belgium, July 4–8, 2016 p1.054

    [23]

    Stangeby P C, Sang C F 2017 Nucl. Fusion 57 056007Google Scholar

    [24]

    马逐曦 2023 博士学位论文 (长沙: 中南大学)

    Ma Z X 2023 PhD Dissertation (Changsha: Central South University

    [25]

    Meha D, Manan S 2021 Clin. eHealth 4 1Google Scholar

    [26]

    Naskath J, Sivakamasundari G, Begum A A S 2022 Wirel. Pers. Commun. 128 2913

    [27]

    Glorot X, Bordes A, Bengio Y 2011 The 14th International Conference on Artificial Intelligence and Statistics Fort Lauderdale, FL, USA, April 11–13, 2011 p315

    [28]

    Nair V, Hinton G E 2010 The 27th International Conference on Machine Learning (ICML) Haifa, Israel, June 21—24, 2010 p807

    [29]

    Hochreiter S 1998 Int. J. Uncertain. Fuzziness Knowl. Based Syst. 6 107

    [30]

    Krizhevsky A, Sutskever I, Hinton G E 2017 Commun. ACM 60 84Google Scholar

    [31]

    任进军, 王宁 2018 甘肃高师学报 23 61Google Scholar

    Ren J J, Wang N 2018 Gansu Gaoshi Xuebao 23 61Google Scholar

    [32]

    Touvron H, Bojanowski P, Caron M, Cord M, El-Nouby A, Grave E, Izacard G, Joulin A, Synnaeve G, Verbeek J, Jégou H 2022 IEEE Trans. Pattern Anal. Mach. Intell. 45 5314

  • [1] 郭焱, 吕恒, 丁春玲, 袁晨智, 金锐博. 分数阶涡旋光衍射过程的机器学习识别. 物理学报, doi: 10.7498/aps.74.20241458
    [2] 张童, 王加豪, 田帅, 孙旭冉, 李日. 基于机器学习的铸件凝固过程动态收缩行为. 物理学报, doi: 10.7498/aps.74.20241581
    [3] 王鹏, 麦麦提尼亚孜·麦麦提阿卜杜拉. 机器学习的量子动力学. 物理学报, doi: 10.7498/aps.74.20240999
    [4] 刘宇杰, 王振宇, 雷航, 张国宇, 咸家伟, 高志斌, 孙军, 宋海峰, 丁向东. 卷积神经网络辅助无机晶体弹性性质预测. 物理学报, doi: 10.7498/aps.74.20250127
    [5] 张旭, 丁进敏, 侯晨阳, 赵一鸣, 刘鸿维, 梁生. 基于机器学习的激光匀光整形方法. 物理学报, doi: 10.7498/aps.73.20240747
    [6] 张嘉晖. 蛋白质计算中的机器学习. 物理学报, doi: 10.7498/aps.73.20231618
    [7] 欧阳鑫健, 张岩星, 王之龙, 张锋, 陈韦嘉, 庄园, 揭晓, 刘来君, 王大威. 面向铁电相变的机器学习: 基于图卷积神经网络的分子动力学模拟. 物理学报, doi: 10.7498/aps.73.20240156
    [8] 刘烨, 牛赫然, 李兵兵, 马欣华, 崔树旺. 机器学习在宇宙线粒子鉴别中的应用. 物理学报, doi: 10.7498/aps.72.20230334
    [9] 管星悦, 黄恒焱, 彭华祺, 刘彦航, 李文飞, 王炜. 生物分子模拟中的机器学习方法. 物理学报, doi: 10.7498/aps.72.20231624
    [10] 方波浪, 王建国, 冯国斌. 基于物理信息神经网络的光斑质心计算. 物理学报, doi: 10.7498/aps.71.20220670
    [11] 林键, 叶梦, 朱家纬, 李晓鹏. 机器学习辅助绝热量子算法设计. 物理学报, doi: 10.7498/aps.70.20210831
    [12] 陈江芷, 杨晨温, 任捷. 基于波动与扩散物理系统的机器学习. 物理学报, doi: 10.7498/aps.70.20210879
    [13] 张瑶, 张云波, 陈立. 基于深度学习的光学表面杂质检测. 物理学报, doi: 10.7498/aps.70.20210403
    [14] 王伟, 揭泉林. 基于机器学习J1-J2反铁磁海森伯自旋链相变点的识别方法. 物理学报, doi: 10.7498/aps.70.20210711
    [15] 王鹏举, 范俊宇, 苏艳, 赵纪军. 基于机器学习构建的环三亚甲基三硝胺晶体势. 物理学报, doi: 10.7498/aps.69.20200690
    [16] 王永生, 孙 瑾, 王昌金, 范洪达. 变参数混沌时间序列的神经网络预测研究. 物理学报, doi: 10.7498/aps.57.6120
    [17] 王瑞敏, 赵 鸿. 神经元传输函数对人工神经网络动力学特性的影响. 物理学报, doi: 10.7498/aps.56.730
    [18] 王耀南, 谭 文. 混沌系统的遗传神经网络控制. 物理学报, doi: 10.7498/aps.52.2723
    [19] 谭文, 王耀南, 刘祖润, 周少武. 非线性系统混沌运动的神经网络控制. 物理学报, doi: 10.7498/aps.51.2463
    [20] 神经网络的自适应删剪学习算法及其应用. 物理学报, doi: 10.7498/aps.50.674
计量
  • 文章访问数:  292
  • PDF下载量:  4
  • 被引次数: 0
出版历程
  • 收稿日期:  2025-03-24
  • 修回日期:  2025-04-18
  • 上网日期:  2025-05-10

/

返回文章
返回