Implementing deep learning-based disruption prediction in a drifting data environment of new tokamak: HL-3
A deep learning-based disruption prediction algorithm has been implemented on a new tokamak, HL-3. An Area Under receiver-operator characteristic Curve of 0.940 has been realized offline over a test campaign involving 72 disruptive and 240 non-disruptive shots, despite the limited training data avai...
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/ada396 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1841556242614452224 |
|---|---|
| author | Zongyu Yang Wulyu Zhong Fan Xia Zhe Gao Xiaobo Zhu Jiyuan Li Liwen Hu Zhaohe Xu Da Li Guohui Zheng Yihang Chen Junzhao Zhang Bo Li Xiaolong Zhang Yiren Zhu Ruihai Tong Yunbo Dong Yipo Zhang Boda Yuan Xin Yu Zongyuhui He Wenjing Tian Xinwen Gong Min Xu |
| author_facet | Zongyu Yang Wulyu Zhong Fan Xia Zhe Gao Xiaobo Zhu Jiyuan Li Liwen Hu Zhaohe Xu Da Li Guohui Zheng Yihang Chen Junzhao Zhang Bo Li Xiaolong Zhang Yiren Zhu Ruihai Tong Yunbo Dong Yipo Zhang Boda Yuan Xin Yu Zongyuhui He Wenjing Tian Xinwen Gong Min Xu |
| author_sort | Zongyu Yang |
| collection | DOAJ |
| description | A deep learning-based disruption prediction algorithm has been implemented on a new tokamak, HL-3. An Area Under receiver-operator characteristic Curve of 0.940 has been realized offline over a test campaign involving 72 disruptive and 240 non-disruptive shots, despite the limited training data available from the initial two campaigns. In addition to the well-documented challenge of insufficient training data, a previously unanticipated issue is addressed that the data distribution of a new device is continuously drifting. The plasma scans across a broad parameter space, bringing a drifting distribution of disruption causes and diagnostic data. This problem is often overlooked in previous implementations on steadily operating tokamaks, necessitating greater attention in future tokamaks like ITER. To address these challenges, innovative modules including predict-first neural network, data augmentation, and pseudo data placeholders are developed and implemented, which promotes the accuracy by up to 20%. A series of advantages are also brought by the modules, including the robustness in handling missing input channels, and the interpretability to identify which parameter of plasma is under abnormal condition. The results demonstrate that, with dedicated data collection and algorithm implementation, the issues of limited data and drifting distribution can be overcome, and further, the deep learning-based algorithm can reliably provide disruption alarms on a new tokamak. |
| format | Article |
| id | doaj-art-bfd9c635938c4bf2b5fdc047f3c8223c |
| institution | Kabale University |
| issn | 0029-5515 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | IOP Publishing |
| record_format | Article |
| series | Nuclear Fusion |
| spelling | doaj-art-bfd9c635938c4bf2b5fdc047f3c8223c2025-01-07T12:12:32ZengIOP PublishingNuclear Fusion0029-55152025-01-0165202603010.1088/1741-4326/ada396Implementing deep learning-based disruption prediction in a drifting data environment of new tokamak: HL-3Zongyu Yang0https://orcid.org/0009-0000-4083-1552Wulyu Zhong1https://orcid.org/0000-0001-8217-9400Fan Xia2Zhe Gao3https://orcid.org/0000-0003-0275-6330Xiaobo Zhu4https://orcid.org/0009-0004-2762-7038Jiyuan Li5Liwen Hu6Zhaohe Xu7Da Li8Guohui Zheng9Yihang Chen10Junzhao Zhang11Bo Li12Xiaolong Zhang13Yiren Zhu14Ruihai Tong15Yunbo Dong16Yipo Zhang17Boda Yuan18Xin Yu19Zongyuhui He20Wenjing Tian21https://orcid.org/0009-0009-3172-3512Xinwen Gong22Min Xu23https://orcid.org/0009-0001-3059-7026Southwestern Institute of Physic , Chengdu 610041, ChinaSouthwestern Institute of Physic , Chengdu 610041, ChinaSouthwestern Institute of Physic , Chengdu 610041, ChinaDepartment of Engineering Physics, Tsinghua University , Beijing 100084, ChinaSouthwestern Institute of Physic , Chengdu 610041, ChinaSouthwestern Institute of Physic , Chengdu 610041, ChinaSouthwestern Institute of Physic , Chengdu 610041, ChinaSouthwestern Institute of Physic , Chengdu 610041, ChinaSouthwestern Institute of Physic , Chengdu 610041, ChinaSouthwestern Institute of Physic , Chengdu 610041, China; School of Physics, Nankai University , Tianjin 300071, ChinaSouthwestern Institute of Physic , Chengdu 610041, ChinaSouthwestern Institute of Physic , Chengdu 610041, ChinaSouthwestern Institute of Physic , Chengdu 610041, ChinaSouthwestern Institute of Physic , Chengdu 610041, ChinaSouthwestern Institute of Physic , Chengdu 610041, ChinaSouthwestern Institute of Physic , Chengdu 610041, ChinaSouthwestern Institute of Physic , Chengdu 610041, ChinaSouthwestern Institute of Physic , Chengdu 610041, ChinaSouthwestern Institute of Physic , Chengdu 610041, ChinaSouthwestern Institute of Physic , Chengdu 610041, ChinaSouthwestern Institute of Physic , Chengdu 610041, ChinaSouthwestern Institute of Physic , Chengdu 610041, China; Department of Engineering Physics, Tsinghua University , Beijing 100084, ChinaSouthwestern Institute of Physic , Chengdu 610041, ChinaSouthwestern Institute of Physic , Chengdu 610041, ChinaA deep learning-based disruption prediction algorithm has been implemented on a new tokamak, HL-3. An Area Under receiver-operator characteristic Curve of 0.940 has been realized offline over a test campaign involving 72 disruptive and 240 non-disruptive shots, despite the limited training data available from the initial two campaigns. In addition to the well-documented challenge of insufficient training data, a previously unanticipated issue is addressed that the data distribution of a new device is continuously drifting. The plasma scans across a broad parameter space, bringing a drifting distribution of disruption causes and diagnostic data. This problem is often overlooked in previous implementations on steadily operating tokamaks, necessitating greater attention in future tokamaks like ITER. To address these challenges, innovative modules including predict-first neural network, data augmentation, and pseudo data placeholders are developed and implemented, which promotes the accuracy by up to 20%. A series of advantages are also brought by the modules, including the robustness in handling missing input channels, and the interpretability to identify which parameter of plasma is under abnormal condition. The results demonstrate that, with dedicated data collection and algorithm implementation, the issues of limited data and drifting distribution can be overcome, and further, the deep learning-based algorithm can reliably provide disruption alarms on a new tokamak.https://doi.org/10.1088/1741-4326/ada396disruption predictiondeep learningHL-3tokamak |
| spellingShingle | Zongyu Yang Wulyu Zhong Fan Xia Zhe Gao Xiaobo Zhu Jiyuan Li Liwen Hu Zhaohe Xu Da Li Guohui Zheng Yihang Chen Junzhao Zhang Bo Li Xiaolong Zhang Yiren Zhu Ruihai Tong Yunbo Dong Yipo Zhang Boda Yuan Xin Yu Zongyuhui He Wenjing Tian Xinwen Gong Min Xu Implementing deep learning-based disruption prediction in a drifting data environment of new tokamak: HL-3 Nuclear Fusion disruption prediction deep learning HL-3 tokamak |
| title | Implementing deep learning-based disruption prediction in a drifting data environment of new tokamak: HL-3 |
| title_full | Implementing deep learning-based disruption prediction in a drifting data environment of new tokamak: HL-3 |
| title_fullStr | Implementing deep learning-based disruption prediction in a drifting data environment of new tokamak: HL-3 |
| title_full_unstemmed | Implementing deep learning-based disruption prediction in a drifting data environment of new tokamak: HL-3 |
| title_short | Implementing deep learning-based disruption prediction in a drifting data environment of new tokamak: HL-3 |
| title_sort | implementing deep learning based disruption prediction in a drifting data environment of new tokamak hl 3 |
| topic | disruption prediction deep learning HL-3 tokamak |
| url | https://doi.org/10.1088/1741-4326/ada396 |
| work_keys_str_mv | AT zongyuyang implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT wulyuzhong implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT fanxia implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT zhegao implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT xiaobozhu implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT jiyuanli implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT liwenhu implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT zhaohexu implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT dali implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT guohuizheng implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT yihangchen implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT junzhaozhang implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT boli implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT xiaolongzhang implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT yirenzhu implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT ruihaitong implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT yunbodong implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT yipozhang implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT bodayuan implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT xinyu implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT zongyuhuihe implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT wenjingtian implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT xinwengong implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 AT minxu implementingdeeplearningbaseddisruptionpredictioninadriftingdataenvironmentofnewtokamakhl3 |