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 ›› 2023, Vol. 43 ›› Issue (2): 191-198.DOI: 10.16568/j.0254-6086.202302012

• 核聚变工程 • 上一篇    下一篇

基于神经网络的EAST-NBI实验数据分析系统研究

刘智民1, 2,汪金新1, 2,赵远哲*1,谢亚红1, 2,谢远来1, 2   

  1. (1. 中国科学院等离子体物理研究所,合肥 230031; 2. 中国科学技术大学,合肥 230026)
  • 收稿日期:2021-05-18 修回日期:2022-10-22 出版日期:2023-06-15 发布日期:2023-06-12
  • 作者简介:刘智民(1962−),男,安徽合肥人,博士,研究员,博士生导师,从事核聚变实验装置中性束注入系统长脉 冲高功率电源技术研究和实验运行。
  • 基金资助:
    国家重点研发计划(2017YFE0300503) 

Research on EAST-NBI experimental data analysis system based on neural networks 

LIU Zhi-min 1, 2, WANG Jin-xin 1, 2, ZHAO Yuan-zhe 1 , XIE Ya-hong 1, 2, XIE Yuan-lai1, 2   

  1. (1. Institute of Plasma Physics, Chinese Academy of Sciences, Hefei 230031; 2. University of Science and Technology of China, Hefei 230026) 
  • Received:2021-05-18 Revised:2022-10-22 Online:2023-06-15 Published:2023-06-12

摘要: NBI 离子源的运行参数不匹配会导致离子源中的等离子体不稳定,并可能导致离子源打火,从而限 制了 NBI 长脉冲和高功率的运行。为了稳定 NBI 离子源中的等离子体,使用来自多轮实验的数据以及从预测性等离子体模型获得的先验信息,针对现有的实验数据值提取算法精度不高的问题,采用 OPTICS 聚类算法提高了实验数据提取的准确性,为实验操作过程中的故障分析提供参考。同时采用了一种基于自组织特征映射图(SOM) 和逆向传播(BP)神经网络的方法,通过训练历史数据,在给定参数下估算 NBI 离子源束引出过程中的脉冲宽度, 以调整运行参数。 

关键词: NBI, 离子源, OPTICS, SOM, BP 神经网络

Abstract: The mismatch of the operation parameters of the NBI(neutral beam injection) ion source of the nuclear fusion experimental device will lead to the instability of the plasma in the ion source, and maybe cause breakdown in the ion source, which will limit the long pulse and high power operation of the NBI. To stabilize the plasma in the ion source, data from multiple campaigns of experiments along with a prior information obtained from a predictive plasma model are used. For the problem of poor accuracy of the existing key-values extraction algorithms for experimental data, the OPTICS (Ordering Point To Identify the Cluster Structure) algorithm is used to improve the accuracy of the experimental data extraction, and provide a reference for the failure analysis during the experimental operation. A method based on Self-Organizing feature Map (SOM) and Back Propagation (BP) neural network is adopted. By training historical data, the pulse width of the NBI ion source beam extraction process is estimated under given parameters to adjust the operating parameters. 

Key words: NBI, Ion Source, OPTICS, SOM, BP Neural network

中图分类号: