Diagnostic-signals-based identification and AI-prediction for early warning of tearing modes in tokamak
Diagnostic-signals-based identification and AI-prediction for early warning of tearing modes in tokamak
ShengYiWang 1Ruijie Yin 1Wei Zhang 2Baofeng Gao 3Xiaofei Zhao1
作者信息
- 1. School of Mathematics and Statistics & Computational Sciences Hubei Key Laboratory, Wuhan University, 430072 Wuhan, China
- 2. Institute of Plasma Physics, Hefei Institutes of Physical Science, Chinese Academy of Sciences, 230031 Hefei, China
- 3. Key Laboratory of Computing Power Network and Information Security, Ministry of Education, Shandong Computer Science Center (National Supercomputer Center in Jinan), Qilu University of Technology (Shandong Academy of Sciences), 250103 Jinan, People’s Republic of China
- 折叠
摘要
Tearing modes (TMs) are important magnetohydrodynamic instabilities in tokamaks and may serve as precursors to major disruptions. Accurate identification and early warning of TMs are therefore essential for real-time monitoring and control. In this work, we propose a diagnostic-signal-based framework that combines DSDetect for TMs identification and DSPred for feature prediction. DSDetect provides offline and online identification methods to determine TMs warning states, toroidal mode number, and radial location from magnetic probe and electron cyclotron emission signals. Evaluations on EAST experimental data reveal that the online method yields over 97% agreement with the offline reference. DSPred can efficiently forecast key TMs-related feature signals. By subsequently applying DSDetect to these predicted features, the framework effectively generates predictive warning and identification results. Detailed tests and analyses were conducted on three typical discharges exhibiting TMs. The results demonstrate that the proposed method can provide an early-warning lead time exceeding 20 ms, positioning it as a promising candidate for future real-time control applications.
Abstract
Tearing modes (TMs) are important magnetohydrodynamic instabilities in tokamaks and may serve as precursors to major disruptions. Accurate identification and early warning of TMs are therefore essential for real-time monitoring and control. In this work, we propose a diagnostic-signal-based framework that combines DSDetect for TMs identification and DSPred for feature prediction. DSDetect provides offline and online identification methods to determine TMs warning states, toroidal mode number, and radial location from magnetic probe and electron cyclotron emission signals. Evaluations on EAST experimental data reveal that the online method yields over 97% agreement with the offline reference. DSPred can efficiently forecast key TMs-related feature signals. By subsequently applying DSDetect to these predicted features, the framework effectively generates predictive warning and identification results. Detailed tests and analyses were conducted on three typical discharges exhibiting TMs. The results demonstrate that the proposed method can provide an early-warning lead time exceeding 20 ms, positioning it as a promising candidate for future real-time control applications.关键词
tearing modes/tokamak/machine learningKey words
tearing modes/tokamak/machine learning引用本文复制引用
ShengYiWang,Ruijie Yin,Wei Zhang,Baofeng Gao,Xiaofei Zhao.Diagnostic-signals-based identification and AI-prediction for early warning of tearing modes in tokamak[EB/OL].(2026-09-24)[2026-09-29].https://chinaxiv.org/abs/202609.00355.学科分类
工程基础科学
