Tokamak plasma disruption precursor onset time study based on semi-supervised anomaly detection
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
Plasma disruption in tokamak experiments is a challenging issue that causes damage to the device. Reliable prediction methods are needed, but the lack of full understanding of plasma disruption limits the effectiveness of physics-driven methods. Data-driven methods based on supervised learning are commonly used, and they rely on labelled training data. However, manual labelling of disruption precursors is a time-consuming and challenging task, as some precursors are difficult to accurately identify. The mainstream labelling methods assume that the precursor onset occurs at a fixed time before disruption, which leads to mislabeled samples and suboptimal prediction performance. In this paper, we present disruption prediction methods based on anomaly detection to address these issues, demonstrating good prediction performance on J-TEXT and EAST. By evaluating precursor onset times using different anomaly detection algorithms, it is found that labelling methods can be improved since the onset times of different shots are not necessarily the same. The study optimizes precursor labelling using the onset times inferred by the anomaly detection predictor and test the optimized labels on supervised learning disruption predictors. The results on J-TEXT and EAST show that the models trained on the optimized labels outperform those trained on fixed onset time labels.
Abstract
Plasma disruption in tokamak experiments is a challenging issue that causes damage to the device. Reliable prediction methods are needed, but the lack of full understanding of plasma disruption limits the effectiveness of physics-driven methods. Data-driven methods based on supervised learning are commonly used, and they rely on labelled training data. However, manual labelling of disruption precursors is a time-consuming and challenging task, as some precursors are difficult to accurately identify. The mainstream labelling methods assume that the precursor onset occurs at a fixed time before disruption, which leads to mislabeled samples and suboptimal prediction performance. In this paper, we present disruption prediction methods based on anomaly detection to address these issues, demonstrating good prediction performance on J-TEXT and EAST. By evaluating precursor onset times using different anomaly detection algorithms, it is found that labelling methods can be improved since the onset times of different shots are not necessarily the same. The study optimizes precursor labelling using the onset times inferred by the anomaly detection predictor and test the optimized labels on supervised learning disruption predictors. The results on J-TEXT and EAST show that the models trained on the optimized labels outperform those trained on fixed onset time labels.
참고문헌 (49)
- J. Zheng, "Recent progress in Chinese fusion research based on superconducting tokamak configuration", Innovation, 3(4), pp.13~ (2022)
- F. C. Schuller, "Disruptions in tokamaks", Plasma Phys. Contr. Fusion, 37(11A), pp.A135~A162 (1995)
- V. Riccardo, "Disruptions and disruption mitigation", Plasma Phys. Contr. Fusion, 45(12A), pp.A269~A284 (2003)
- P. C. De Vries, "Survey of disruption causes at JET", Nucl. Fusion, 51(5), pp.12~ (2011)
- G. Pautasso, "The ITER disruption mitigation trigger: developing its preliminary design", Nucl. Fusion, 58(3), pp.036011~ (2018)
- G. Sias, "A locked mode indicator for disruption prediction on JET and ASDEX upgrade", Fusion Eng. Des, 138, pp.254~266 (2018)
- B. Cannas, "Support vector machines for disruption prediction and novelty detection at JET", Fusion Eng. Des, 82(5–14), pp.1124~1130 (2006)
- J. M. Lopez, "Implementation of the disruption predictor APODIS in JET’s realtime network using the MARTe framework", IEEE Trans. Nucl. Sci, 61(2), pp.741~744 (2014)
- R. A. Tinguely, "An application of survival analysis to disruption prediction via Random Forests", Plasma Phys. Contr. Fusion, 61(9), pp.095009~ (2019)
- W. Hu, "Real-time prediction of high density EAST disruptions using Random Forest", Nucl. Fusion, 60(6), pp.066034~ (2021)
- D. Wroblewski, "Tokamak disruption alarm based on a neural network model of the high-beta limit", Nucl. Fusion, 37(6), pp.725~ (1997)
- A. Sengupta, "Prediction of density limit disruption boundaries from diagnostic signals using neural networks", Nucl. Fusion, 41(5), pp.487~501 (2001)
- C. G. Windsor, "A cross-tokamak neural network disruption predictor for the JET and ASDEX upgrade tokamaks", Nucl. Fusion, 45(5), pp.337~350 (2005)
- Z. Yang, "A disruption predictor based on a 1.5-dimensional convolutional neural network in HL-2A", Nucl. Fusion, 60(1), pp.016017~ (2019)
- K. J. Montes, "Machine learning for disruption warnings on Alcator C-Mod, DIII-D, and EAST", Nucl. Fusion, 59(9), pp.096015~ (2019)
- B. H. Guo, "Disruption prediction on EAST tokamak using a deep learning algorithm", Plasma Phys. Contr. Fusion, 63(11), pp.115007~ (2021)
- Y. Zhong, "Disruption prediction and model analysis using LightGBM on JTEXT and HL-2A", Plasma Phys. Contr. Fusion, 63(7), pp.075008~ (2021)
- Z. Yang, "In-depth research on the interpretable disruption predictor in HL-2A", Nucl. Fusion, 61(12), pp.126042~ (2021)
- B. Cannas, "A prediction tool for real-time application in the disruption protection system at JET", Nucl. Fusion, 47(11), pp.1559~ (2007)
- B. Cannas, "Disruption forecasting at JET using neural networks", Nucl. Fusion, 44(1), pp.68~76 (2003)
- J. Vega, "Adaptive high learning rate probabilistic disruption predictors from scratch for the next generation of tokamaks", Nucl. Fusion, 54(12), pp.123001~ (2014)
- C. Rea, "Exploratory machine learning studies for disruption prediction using large databases on DIII-D", Fusion Sci. Technol, 74(1–2), pp.89~100 (2018)
- C. Rea, "Disruption prediction investigations using Machine Learning tools on DIII-D and Alcator C-Mod", Plasma Phys. Contr. Fusion, 60(8), pp.084004~ (2018)
- G. Pautasso, "On-line prediction and mitigation of disruptions in ASDEX Upgrade", Nucl. Fusion, 42(1), pp.100~ (2002)
- B. Cannas, "An adaptive real-time disruption predictor for ASDEX upgrade", Nucl. Fusion, 50(7), pp.075004~ (2010)
- Y. Zhang, "Prediction of disruptions on ASDEX Upgrade using discriminant analysis", Nucl. Fusion, 51(6), pp.063039~ (2011)
- S. P. Gerhardt, "Detection of disruptions in the high-β spherical torus NSTX", Nucl. Fusion, 53(6), pp.063021~ (2013)
- R. Yoshino, "Neural-net disruption predictor in JT-60U", Nucl. Fusion, 43(12), pp.1771~1786 (2003)
- R. Yoshino, "Neural-net predictor for beta limit disruptions in JT-60U", Nucl. Fusion, 45(11), pp.1232~1246 (2005)
- J. Jang, "Feature concentration for supervised and semi-supervised learning with unbalanced datasets in visual inspection", IEEE Trans. Ind. Electron, 68(8), pp.7620~7630 (2020)
- M. H. Nguyen, "Impacts of unbalanced test data on the evaluation of classification methods", Int. J. Adv. Comput. Sci. Appl, 10(3), pp.497~502 (2019)
- B. Cannas, "An adaptive real-time disruption predictor for ASDEX Upgrade", Nucl. Fusion, 50(7), pp.075004~ (2010)
- J. Zhu, "Hybrid deep-learning architecture for general disruption prediction across multiple tokamaks", Nucl. Fusion, 61(2), pp.026007~ (2021)
- A. Pau, "A machine learning approach based on generative topographic mapping for disruption prevention and avoidance at JET", Nucl. Fusion, 59(10), pp.106017~ (2019)
- Y. Xiong, "Recognizing multivariate geochemical anomalies for mineral exploration by combining deep learning and one-class support vector machine", Comput. Geosci, 140, pp.104484~ (2020)
- W. Zheng, "Disruption predictor based on neural network and anomaly detection on J-TEXT", Plasma Phys. Contr. Fusion, 62(4), pp.045012~ (2020)
- S. Esquembri, "Real-time Implementation in JET of the SPAD disruption predictor using MARTe", IEEE Trans. Nucl. Sci, 65(2), pp.836~842 (2018)
- R. Aledda, "Multivariate statistical models for disruption prediction at ASDEX upgrade", Fusion Eng. Des, 88(6–8), pp.1297~1301 (2013)
- I. Goodfellow, "Deep learning 1", pp.499~507 (2016), MIT press
- G. E. Hinton, "Autoencoders, minimum description length and Helmholtz free energy", Advances in Neural Information Processing Systems San Mateo, 6 (1994)
- Scholkopf, "Estimating the Support of a High-Dimensional Distribution", Neural Computation, 13(7), pp.1443~1471 (2001)
- M. Y. Su, "Real-time anomaly detection systems for Denial-of-Service attacks by weighted k-nearest-neighbor classifiers", Expert Syst. Appl, 38(4), pp.3492~3498 (2011)
- S. Esquembri, "Real-time implementation in JET of the SPAD disruption predictor using MARTe", IEEE Trans. Nucl. Sci, 65(2), pp.836~842 (2018)
- H. P. Kriegel, "Angle-based Outlier Detection in High-Dimensional Data", Proceedings of the 14th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (2008), ACM
- C. S. Shen, "IDP-PGFE : an interpretable disruption predictor based on physicsguided feature extraction", Nucl. Fusion, 63(4), pp.046024~ (2023)
- Y. H. Ding, "Overview of the J-TEXT progress on RMP and disruption physics", Plasma Sci. Technol, 20(12), pp.125101~ (2018)
- Y. Liang, "Overview of the recent experimental research on the J-TEXT tokamak", Nucl. Fusion, 59(11), pp.112016~ (2019)
- D. L. Chen, "Characterization of disruption halo current between ‘w-like’graphite divertor and “ITER-like” divertor structure on east tokamak", Plasma Phys. Contr. Fusion, 62(9), pp.095019~ (2020)
- X. Gao, "Experimental progress of hybrid operational scenario on EAST tokamak", Nucl. Fusion, 60(10), pp.102001~ (2020)