Adaptive anomaly detection disruption prediction starting from first discharge on tokamak
Plasma disruption presents a significant challenge in tokamak fusion, especially in large-size devices like ITER, where it causes severe damage. While current data-driven machine learning methods perform well in disruption prediction, they require extensive discharge data for model training. However...
Saved in:
| Main Authors: | , , , , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IOP Publishing
2025-01-01
|
| Series: | Nuclear Fusion |
| Subjects: | |
| Online Access: | https://doi.org/10.1088/1741-4326/ada9a9 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1823858104317509632 |
|---|---|
| author | X.K. Ai W. Zheng M. Zhang Y.H. Ding D.L. Chen Z.Y. Chen B.H. Guo C.S. Shen N.C. Wang Z.J. Yang Z.P. Chen Y. Pan B. Shen B.J. Xiao J-TEXT team |
| author_facet | X.K. Ai W. Zheng M. Zhang Y.H. Ding D.L. Chen Z.Y. Chen B.H. Guo C.S. Shen N.C. Wang Z.J. Yang Z.P. Chen Y. Pan B. Shen B.J. Xiao J-TEXT team |
| author_sort | X.K. Ai |
| collection | DOAJ |
| description | Plasma disruption presents a significant challenge in tokamak fusion, especially in large-size devices like ITER, where it causes severe damage. While current data-driven machine learning methods perform well in disruption prediction, they require extensive discharge data for model training. However, future tokamaks will begin operations without any prior data, making it difficult to train data-driven disruption predictors and select appropriate hyperparameters during the early operation period. In this period disruption prediction also aims to support safe exploration of operation range and accumulate necessary data to develop advanced prediction models. Thus, predictors must adapt to evolving plasma states during this exploration phase. To address these challenges, this study further develops the enhanced convolutional autoencoder anomaly detection (E-CAAD) predictor and proposes a cross-tokamak adaptive transfer method based on E-CAAD. By training the E-CAAD model on data from existing devices, the predictor can effectively distinguish between disruption precursor and non-disruption samples in new device, enabling disruption prediction from the first shot on the new device. Additionally, adaptive learning from scratch and alarm threshold adaptive adjustment strategies are proposed to enable model automatically adapt to changes in the discharge scenario. The adaptive learning strategy enables the predictor to fully use scarce data during the early operation of the new device while rapidly adapting to changes in the discharge scenario. The threshold adaptive adjustment strategy addresses the challenge of selecting alarm thresholds on new devices where the validation set is lacking, ensuring that the alarm thresholds adapt to changes in the discharge scenario. Finally, the experiment transferring the model from J-TEXT to EAST exhibit that this method enables disruption prediction from the first shot on EAST, allowing the predictor to adapt to changes in the discharge scenario and maintain high prediction performance. |
| format | Article |
| id | doaj-art-298ce0e241674d55ac2d03eeb4c9b38f |
| institution | Kabale University |
| issn | 0029-5515 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | IOP Publishing |
| record_format | Article |
| series | Nuclear Fusion |
| spelling | doaj-art-298ce0e241674d55ac2d03eeb4c9b38f2025-02-11T14:09:48ZengIOP PublishingNuclear Fusion0029-55152025-01-0165303601110.1088/1741-4326/ada9a9Adaptive anomaly detection disruption prediction starting from first discharge on tokamakX.K. Ai0https://orcid.org/0009-0004-5599-1953W. Zheng1https://orcid.org/0000-0002-2853-6021M. Zhang2https://orcid.org/0000-0002-9372-4926Y.H. Ding3D.L. Chen4https://orcid.org/0000-0001-7093-3154Z.Y. Chen5https://orcid.org/0000-0002-8934-0364B.H. Guo6https://orcid.org/0000-0003-2630-5796C.S. Shen7https://orcid.org/0000-0003-3503-8140N.C. Wang8https://orcid.org/0000-0001-6797-2398Z.J. Yang9https://orcid.org/0000-0002-9141-7869Z.P. Chen10https://orcid.org/0000-0002-8330-0070Y. Pan11B. Shen12B.J. Xiao13J-TEXT team14State Key Laboratory of Advanced Electromagnetic Technology, International Joint Research Laboratory of Magnetic Confinement Fusion and Plasma Physics, School of Electrical and Electronic Engineering, Huazhong University of Science and Technology , Wuhan 430074, ChinaState Key Laboratory of Advanced Electromagnetic Technology, International Joint Research Laboratory of Magnetic Confinement Fusion and Plasma Physics, School of Electrical and Electronic Engineering, Huazhong University of Science and Technology , Wuhan 430074, ChinaState Key Laboratory of Advanced Electromagnetic Technology, International Joint Research Laboratory of Magnetic Confinement Fusion and Plasma Physics, School of Electrical and Electronic Engineering, Huazhong University of Science and Technology , Wuhan 430074, ChinaState Key Laboratory of Advanced Electromagnetic Technology, International Joint Research Laboratory of Magnetic Confinement Fusion and Plasma Physics, School of Electrical and Electronic Engineering, Huazhong University of Science and Technology , Wuhan 430074, ChinaInstitute of Plasma Physics , Chinese Academy of Sciences, Hefei 230031, ChinaState Key Laboratory of Advanced Electromagnetic Technology, International Joint Research Laboratory of Magnetic Confinement Fusion and Plasma Physics, School of Electrical and Electronic Engineering, Huazhong University of Science and Technology , Wuhan 430074, ChinaInstitute of Plasma Physics , Chinese Academy of Sciences, Hefei 230031, ChinaState Key Laboratory of Advanced Electromagnetic Technology, International Joint Research Laboratory of Magnetic Confinement Fusion and Plasma Physics, School of Electrical and Electronic Engineering, Huazhong University of Science and Technology , Wuhan 430074, ChinaState Key Laboratory of Advanced Electromagnetic Technology, International Joint Research Laboratory of Magnetic Confinement Fusion and Plasma Physics, School of Electrical and Electronic Engineering, Huazhong University of Science and Technology , Wuhan 430074, ChinaState Key Laboratory of Advanced Electromagnetic Technology, International Joint Research Laboratory of Magnetic Confinement Fusion and Plasma Physics, School of Electrical and Electronic Engineering, Huazhong University of Science and Technology , Wuhan 430074, ChinaState Key Laboratory of Advanced Electromagnetic Technology, International Joint Research Laboratory of Magnetic Confinement Fusion and Plasma Physics, School of Electrical and Electronic Engineering, Huazhong University of Science and Technology , Wuhan 430074, ChinaState Key Laboratory of Advanced Electromagnetic Technology, International Joint Research Laboratory of Magnetic Confinement Fusion and Plasma Physics, School of Electrical and Electronic Engineering, Huazhong University of Science and Technology , Wuhan 430074, ChinaInstitute of Plasma Physics , Chinese Academy of Sciences, Hefei 230031, ChinaInstitute of Plasma Physics , Chinese Academy of Sciences, Hefei 230031, ChinaState Key Laboratory of Advanced Electromagnetic Technology, International Joint Research Laboratory of Magnetic Confinement Fusion and Plasma Physics, School of Electrical and Electronic Engineering, Huazhong University of Science and Technology , Wuhan 430074, ChinaPlasma disruption presents a significant challenge in tokamak fusion, especially in large-size devices like ITER, where it causes severe damage. While current data-driven machine learning methods perform well in disruption prediction, they require extensive discharge data for model training. However, future tokamaks will begin operations without any prior data, making it difficult to train data-driven disruption predictors and select appropriate hyperparameters during the early operation period. In this period disruption prediction also aims to support safe exploration of operation range and accumulate necessary data to develop advanced prediction models. Thus, predictors must adapt to evolving plasma states during this exploration phase. To address these challenges, this study further develops the enhanced convolutional autoencoder anomaly detection (E-CAAD) predictor and proposes a cross-tokamak adaptive transfer method based on E-CAAD. By training the E-CAAD model on data from existing devices, the predictor can effectively distinguish between disruption precursor and non-disruption samples in new device, enabling disruption prediction from the first shot on the new device. Additionally, adaptive learning from scratch and alarm threshold adaptive adjustment strategies are proposed to enable model automatically adapt to changes in the discharge scenario. The adaptive learning strategy enables the predictor to fully use scarce data during the early operation of the new device while rapidly adapting to changes in the discharge scenario. The threshold adaptive adjustment strategy addresses the challenge of selecting alarm thresholds on new devices where the validation set is lacking, ensuring that the alarm thresholds adapt to changes in the discharge scenario. Finally, the experiment transferring the model from J-TEXT to EAST exhibit that this method enables disruption prediction from the first shot on EAST, allowing the predictor to adapt to changes in the discharge scenario and maintain high prediction performance.https://doi.org/10.1088/1741-4326/ada9a9tokamakdisruption predictiondeep learningadaptive learninganomaly detection |
| spellingShingle | X.K. Ai W. Zheng M. Zhang Y.H. Ding D.L. Chen Z.Y. Chen B.H. Guo C.S. Shen N.C. Wang Z.J. Yang Z.P. Chen Y. Pan B. Shen B.J. Xiao J-TEXT team Adaptive anomaly detection disruption prediction starting from first discharge on tokamak Nuclear Fusion tokamak disruption prediction deep learning adaptive learning anomaly detection |
| title | Adaptive anomaly detection disruption prediction starting from first discharge on tokamak |
| title_full | Adaptive anomaly detection disruption prediction starting from first discharge on tokamak |
| title_fullStr | Adaptive anomaly detection disruption prediction starting from first discharge on tokamak |
| title_full_unstemmed | Adaptive anomaly detection disruption prediction starting from first discharge on tokamak |
| title_short | Adaptive anomaly detection disruption prediction starting from first discharge on tokamak |
| title_sort | adaptive anomaly detection disruption prediction starting from first discharge on tokamak |
| topic | tokamak disruption prediction deep learning adaptive learning anomaly detection |
| url | https://doi.org/10.1088/1741-4326/ada9a9 |
| work_keys_str_mv | AT xkai adaptiveanomalydetectiondisruptionpredictionstartingfromfirstdischargeontokamak AT wzheng adaptiveanomalydetectiondisruptionpredictionstartingfromfirstdischargeontokamak AT mzhang adaptiveanomalydetectiondisruptionpredictionstartingfromfirstdischargeontokamak AT yhding adaptiveanomalydetectiondisruptionpredictionstartingfromfirstdischargeontokamak AT dlchen adaptiveanomalydetectiondisruptionpredictionstartingfromfirstdischargeontokamak AT zychen adaptiveanomalydetectiondisruptionpredictionstartingfromfirstdischargeontokamak AT bhguo adaptiveanomalydetectiondisruptionpredictionstartingfromfirstdischargeontokamak AT csshen adaptiveanomalydetectiondisruptionpredictionstartingfromfirstdischargeontokamak AT ncwang adaptiveanomalydetectiondisruptionpredictionstartingfromfirstdischargeontokamak AT zjyang adaptiveanomalydetectiondisruptionpredictionstartingfromfirstdischargeontokamak AT zpchen adaptiveanomalydetectiondisruptionpredictionstartingfromfirstdischargeontokamak AT ypan adaptiveanomalydetectiondisruptionpredictionstartingfromfirstdischargeontokamak AT bshen adaptiveanomalydetectiondisruptionpredictionstartingfromfirstdischargeontokamak AT bjxiao adaptiveanomalydetectiondisruptionpredictionstartingfromfirstdischargeontokamak AT jtextteam adaptiveanomalydetectiondisruptionpredictionstartingfromfirstdischargeontokamak |