| PHYSICS OF GASES, PLASMAS, AND ELECTRIC DISCHARGES |
Prev
Next
|
|
|
Interpretable logistic regression for predicting 1/1 kink-driven regular complete sawtooth phenomenon in EAST tokamak plasmas |
| Haipeng Quan(全海鹏)1,2,†, Liqing Xu(徐立清)2, Chaowei Mai(麦晁玮)3, Yan Chao(晁燕)2,‡, Yuhang Wang(王宇航)4, and EAST Team2 |
1 School of Computer Science and Engineering, Anhui University of Science and Technology, Huainan 232001, China;
2 Plasma Institute, Hefei Institutes of Physical Science, Chinese Academy of Sciences, Hefei 230031, China;
3 Guangdong Ocean University, School of Electronics and Information Engineering, Zhanjiang 524088, China;
4 School of Physical Science and Information Technology, Anhui University, Hefei 230601, China |
|
|
|
|
Abstract Sawtooth phenomena are a central topic in tokamak fusion research; nevertheless, different sawtooth classes differ markedly in underlying physics, statistical abundance and diagnostic definition. This paper focuses exclusively on regular complete sawtooth — ideal, 1/1 internal-kink-driven events with full magnetic reconnection — which are the most frequent, unambiguously identifiable and theoretically best characterized (hereafter, all references to "sawtooth" denote this specific class). Compound crashes, partial-reconnection events and fast-ion-induced giant sawtooth are deliberately excluded because of their limited data volume and still-contested classification criteria, the inclusion of which would introduce intolerable label noise and theoretical ambiguity. To this aim, we propose a machine-learning-based temporal binary-classification framework that converts multi-diagnostic, high-resolution signals into precise labels distinguishing "oscillation" from "quiet" windows for this specific category, thereby supplying critical timing information for active control. A large-scale database covering 124 discharges and hundreds of millions of samples was constructed, and three representative algorithms — logistic regression, decision tree and random forest — were trained and compared. Among them, logistic regression achieved the best and most robust performance, reaching 95 % accuracy on an independent test set and significantly outperforming the other models. Furthermore, shapley additive explanations (SHAP) was innovatively employed to quantify the contribution magnitude and direction of key physical features to the onset of regular 1/1 sawtooth, substantially enhancing model interpretability and physical fidelity. The study provides an efficient and robust predictor for the active intervals of ordinary 1/1 sawtooth; the uncovered correlations between physical drivers and sawtooth behavior lay a solid foundation for deepening the understanding of regular sawtooth evolution and for optimizing control strategies, thereby holding significant promise for improving the operational stability of fusion plasmas.
|
Received: 20 October 2025
Revised: 12 January 2026
Accepted manuscript online: 14 January 2026
|
|
PACS:
|
52.35.Py
|
(Macroinstabilities (hydromagnetic, e.g., kink, fire-hose, mirror, ballooning, tearing, trapped-particle, flute, Rayleigh-Taylor, etc.))
|
| |
52.55.Fa
|
(Tokamaks, spherical tokamaks)
|
| |
52.30.-q
|
(Plasma dynamics and flow)
|
| |
07.05.Mh
|
(Neural networks, fuzzy logic, artificial intelligence)
|
|
| Fund: Project supported by the National Key R&D Program of China (Grant No. 2022YFE03010003) and the National Natural Science Foundation of China (Grant No. 12275309). |
Corresponding Authors:
Haipeng Quan, Yan Chao
E-mail: haipeng.quan@ipp.ac.cn;chaoyan@ipp.ac.cn
|
Cite this article:
Haipeng Quan(全海鹏), Liqing Xu(徐立清), Chaowei Mai(麦晁玮), Yan Chao(晁燕), Yuhang Wang(王宇航), and EAST Team Interpretable logistic regression for predicting 1/1 kink-driven regular complete sawtooth phenomenon in EAST tokamak plasmas 2026 Chin. Phys. B 35 045203
|
[1] Hu Q Y, Xu L Q and Liu D J 2021 Chin. Phys. B 30 035201
[2] Pan S S, Duan Y M, Xu L Q, Chao Y, Zhong G Q, Sun Y W, Sheng H, Liu H Q, Chu Y Q, Lü B, Jin Y F and Hu L Q 2023 Acta Phys. Sin. 72 135203 (in Chinese)
[3] Fan T S, Yuan X, Zhang X, Xie X F, Chen Z J, Peng X Y, Du T F, Hu Z M, Cui Z Q, Chen J X, Li X Q, Zhang G H and Wang X G 2013 Sci. Sin.-Phys. Mech. Astron. 43 1236
[4] Xu M, Xu L Q, Zhao H L, Li Y Y, Zhong G Q, Hao B L, Ma R R, Chen W, Liu H Q, Xu G S, Hu J S, Wan B N and EAST Team 2023 Acta Phys. Sin. 72 215204 (in Chinese)
[5] Bandyopadhyay I, Igochine V, Sauter O, et al. 2025 Nucl. Fusion 65 103001
[6] Chapman I T, Hender T C, Saarelma S, Sharapov S E, Akers R J and Conway N J 2006 Nucl. Fusion 46 1009
[7] Mück A, Goodman T P, Maraschek M, Pereverez G, Ryter F and Zohm H 2005 Plasma Phys. Control. Fusion 47 1633
[8] Chapman I T 2011 Plasma Phys. Control. Fusion 53 013001
[9] Felici F, Goodman T P, Sauter O, Canal G, Duval B P, Rossel J X and the TCV Team 2012 Nucl. Fusion 52 074001
[10] Girardo J B, Sharapov S, Boom J, de Vries P, Fitzgerald M, Hawkes N, Kiptily V, Mantsinen M, Nave M F F, Schneider M and JET Contributors 2016 Phys. Plasmas 23 012505
[11] Wen X D, Xu L Q, Hu L Q, Liu H Q, Duan Y M, Chu Y Q, Zhong G Q, Zhang W, Zhang X J and Mai C W 2023 Nucl. Techn. 46 010502
[12] Li J C, Ding Y H, Yu Q Q, Wang N C, Li D, Zhang X Q, Han D L, Chen Z P, Yang Z J, Zhou S, Yan W, Liang Y F, Zhang X L, Lin X D, Sun H B, Gao X and Li J G 2020 Nucl. Fusion 60 126002
[13] Gude A, Maraschek M, Eulenberg P, Igochine V and ASDEX Upgrade Team, Max Planck Institute for Plasma Physics, Max Planck Society 2015 Automated sawtooth detection with multi-signal analysis In 42nd EPS Conference on Plasma Physics (Geneva: European Physical Society)
[14] van den Brand H, de Baar M R, van Berkel M, Blanken T C, Felici F, Westerhof E, Willensdorfer M, ASDEX Upgrade Team and EUROfusion MST1 Team 2016 Plasma Phys. Control. Fusion 58 075002
[15] Isel N, Isayama A, Ishida S, Sato M, Oikawa T, Fukuda T, Nagashima A, Iwama N and JT-60 Team 2001 Fusion Eng. Des. 53 213
[16] Lan G, De Carlo M, van Diggelen F, Tomczak J M, Roijers D M and Eiben A E 2020 arXiv:2001.07804 [cs.NE]
[17] Proulx M J, Eracleous T, Spencer B, Passfield A, de Sousa A and Mohammadi A 2021 arXiv:2105.13295[cs.HC]
[18] Gramacy R B and Polson N G 2012 Bayesian Anal. 7 567
[19] El-Koka A, Cha K H and Kang D K 2013 Regularization parameter tuning optimization approach in logistic regression In 2013 15th International Conference on Advanced Communications Technology (ICACT), PyeongChang, Korea (South), pp. 13-18
[20] Xie Y and Willett R “Online logistic regression on manifolds”, 2013 IEEE International Conference on Acoustics, Speech and Signal Processing, Vancouver, BC, Canada, 2013, pp. 3367-3371
[21] Lundberg SMand Lee S I 2017 Adv. Neural Inf. Process. Syst. 30 4765
[22] Kumar I, Sharma A, Singh S and Kumar A 2022 IEEE Access 10 35421
[23] von Goeler S, StodiekWand SauthoffW1974 Phys. Rev. Lett. 33 1201
[24] Strait E J, DeBoo J C, Hahm T S, Heidbrink W W, Cheng C Z and Jensen T H 1989 Phys. Rev. Lett. 62 741
[25] Liu H Q, Jie Y X, Ding W X, Brower D L, Xu G S, Qian J, Wan B N, Qian J P and EAST Team 2014 Nucl. Fusion 54 023010
[26] Rea C, Granetz R S, Montes K, Tinguely R A, Eidietis N, Hanson J, Meneghini O, Frank S and Sammuli B 2019 Nucl. Fusion 59 056016
[27] Li Y, Liu H, Qian J P, et al. 2021 Nucl. Fusion 61 036014
[28] Zhang B, Xu L, Gong X Z, et al. 2020 Plasma Phys. Control. Fusion 62 045008
[29] Wen X D 2023 Experimental studies on the effect of on-axis ICRF heating on the sawtooth in EAST device [in Chinese] (Master’s Thesis) (Hefei: University of Science and Technology of China) pp. 22-29
[30] Verhulst P J 1838 Corresp. Math. Phys. 10 113 (in French)
[31] Roe B P, Yang H J, Zhu J, Liu Y, Stancu I and McGregor G 2005 Nucl. Instrum. Methods A 543 577
[32] Kirasich K, Smith T and Sadler B 2018 SMU Data Sci. Rev. 1 9 |
| No Suggested Reading articles found! |
|
|
Viewed |
|
|
|
Full text
|
|
|
|
|
Abstract
|
|
|
|
|
Cited |
|
|
|
|
Altmetric
|
|
blogs
Facebook pages
Wikipedia page
Google+ users
|
Online attention
Altmetric calculates a score based on the online attention an article receives. Each coloured thread in the circle represents a different type of online attention. The number in the centre is the Altmetric score. Social media and mainstream news media are the main sources that calculate the score. Reference managers such as Mendeley are also tracked but do not contribute to the score. Older articles often score higher because they have had more time to get noticed. To account for this, Altmetric has included the context data for other articles of a similar age.
View more on Altmetrics
|
|
|
