欢迎访问《核聚变与等离子体物理》官方网站,今天是 分享到:

核聚变与等离子体物理 ›› 2022, Vol. 42 ›› Issue (s1): 164-169.DOI: 10.16568/j.0254-6086.2022s1012

• 等离子体物理 • 上一篇    下一篇

基于卷积神经网络的HL-2A装置H模辨识研究

梁嘉禾1,刘松芬*1,王鸿鹏2,杜 月2,王占辉3,陈逸航3,许 敏3,夏 凡3,杨宗谕3,钟武律3   

  1. (1. 南开大学物理科学学院,天津 300071;2. 南开大学人工智能学院,天津300350;3. 核工业西南物理研究院,成都 610041)
  • 收稿日期:2021-06-28 修回日期:2022-01-17 出版日期:2022-05-15 发布日期:2022-05-24
  • 作者简介:梁嘉禾(2001-),男,吉林长春人,本科,专业方向为物理学专业。
  • 基金资助:
    政府间国际科技创新合作专项磁约束核聚变能发展研究(2017YFE0301702);国家基础科学人才拔尖计划(J11- 03208)

H mode period recognition in HL-2A tokamak based on convolutional neural network

LIANG Jia-he1, LIU Song-fen1, WANG Hong-peng2, DU Yue2, WANG Zhan-hui3,CHEN Yi-hang3, XU Min3, XIA Fan3, YANG Zong-yu3, ZHONG Wu-lü3   

  1. (1. School of Physics, Nankai University, Tianjin 300071; 2. College of Artificial Intelligence, Nankai University, Tianjin 300350; 3. Southwestern Institute of Physics, Chengdu 610041)
  • Received:2021-06-28 Revised:2022-01-17 Online:2022-05-15 Published:2022-05-24

摘要: 基于HL-2A装置的放电实验数据,利用卷积神经网络和时间窗口算法开发了高约束(H)模时段的识别算法,得到了可靠的高成功率的高约束模时段识别结果。算法中,选取206次放电实验数据中等离子体储能及氘a通道信号作为双通道原始数据进行学习,得到一个深度为21层的二分类卷积神经网络。该网络模型经过其他474次放电数据的测试集检验,高约束模识别的正确率达到了98.17%。

关键词: HL-2A装置, 高约束模, 卷积神经网络, 模式识别

Abstract: A machine learning method has been adopted to train models for H-mode prediction from the HL-2A experimental data. Stored energy and deuterium alpha signals from 206 discharges are selected for training, 15 for validation and 474 for testing. H-mode period of all the data is marked previously. Instances of 2-channel time sequence are generated in the way of time window. Finally, a 21 layers convolutional neural network was developed. It scored 98.17% of accuracy in test set and the recognized H-mode beginning and ending time error were lower than 10ms, which can be applied to the high confinement mode research.

Key words: HL-2A tokamak, High confinement mode, Convolutional neural network, Pattern recognition

中图分类号: