Realization of automatic data cleaning and feedback conditioning for J-TEXT ECEI signals based on machine learning

Zhang, Zhichao; Zhi-jian, Yang; Gao, Yuan; Zha, Xianqian; Jin, Ziyang; Luo, Qiang; Zheng, Wei; Zhao, Qing; Wang, Nengchao; Gao, Li; Pan, Yuan

doi:10.1016/j.fusengdes.2022.113065

Cited by 3 publications

(1 citation statement)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…If the attenuation is not properly set up, the pixel channel will suffer from saturated or weak signals and will result in bad images. A module aimed at automatic data cleaning for J-TEXT ECEI signals based on machine learning was developed [33]. A 2-stage classifier model was designed and built, which was able to recognize six types of signal states: low-attenuation, saturated-background, high-background, weak signal, zero signal, and normal signal.…”

Section: Machine Learning In the Diagnostics Data Processingmentioning

confidence: 99%

Overview of machine learning applications in fusion plasma experiments on J-TEXT tokamak

Zheng

Xue²,

Shen³

et al. 2022

Plasma Sci. Technol.

Self Cite

View full text Add to dashboard Cite

Machine learning research and applications in fusion plasma experiments is one of the main subjects on J-TEXT. Since 2013, various kinds of traditional machine learning, as well as deep learning methods have been applied to fusion plasma experiments. Further applications in the real-time experimental environment have proved the feasibility and effectiveness of the methods. For disruption prediction, we started with predicting disruptions of limited classes with a short warning time that was not able to meet the requirements of the mitigation system. With years of study, nowadays disruption prediction methods on J-TEXT are able to predict all kinds of disruptions on J-TEXT with a high success rate and long enough warning time. Furthermore, cross-device disruption prediction methods and interpretable models are studied and have obtained promising results. For diagnostics data processing, efforts have been made to reduce manual work in processing and to increase the robustness of the diagnostic system. Both traditional machine learning and deep learning-based models have been applied to the real-time experimental environment and have been cooperating with plasma control and other systems to make joint decisions to further support the experiments.

show abstract

Section: Machine Learning In the Diagnostics Data Processingmentioning

confidence: 99%

Overview of machine learning applications in fusion plasma experiments on J-TEXT tokamak

Zheng

Xue²,

Shen³

et al. 2022

Plasma Sci. Technol.

Self Cite

View full text Add to dashboard Cite

show abstract

Image inpainting for ECEI based on DeepFillv2 model

Xuan,

Yang,

Lei

et al. 2024

Fusion Engineering and Design

View full text Add to dashboard Cite

Overview of the recent experimental research on the J-TEXT tokamak

Ding,

Wang,

Chen

et al. 2024

Nucl. Fusion

View full text Add to dashboard Cite

The J-TEXT capability is enhanced compared to two years ago with several upgrades of its diagnostics and the increase of electron cyclotron resonance heating (ECRH) power to 1 MW. With the application of electron cyclotron wave (ECW), the ECW assisted plasma startup is achieved; the tearing mode is suppressed; the toroidal injection of 300 kW ECW drives around 24 kA current; fast electrons are generated with toroidal injected ECW and the runaway current conversion efficiency increases with ECRH power. The mode coupling between 2/1 and 3/1 modes are extensively studied. The coupled 2/1 and 3/1 modes usually lead to major disruption. Their coupling can be either suppressed or avoided by external resonant magnetic perturbation (RMP) fields and hence avoids the major disruption. It is also found that the 2/1 threshold of external field is significantly reduced by a pre-excited 3/1 mode, which can be either a locked island or an external kink mode. The disruption control is studied by developing prediction methods capable of cross tokamak application and by new mitigation methods, such as the biased electrode or electromagnetic pellet injector. The high-density operation and related disruptions are studied from various aspects. Approaching the density limit, the collapse of the edge shear layer is observed and such collapse can be prevented by applying edge biasing, leading to an increased density limit. The density limit is also observed to increase, if the plasma is operated in the poloidal divertor configuration or the plasma purity is increased by increasing the pre-filled gas pressure or ECRH power during the start-up phase.

show abstract

Realization of automatic data cleaning and feedback conditioning for J-TEXT ECEI signals based on machine learning

Cited by 3 publications

References 16 publications

Overview of machine learning applications in fusion plasma experiments on J-TEXT tokamak

Overview of machine learning applications in fusion plasma experiments on J-TEXT tokamak

Image inpainting for ECEI based on DeepFillv2 model

Overview of the recent experimental research on the J-TEXT tokamak

Contact Info

Product

Resources

About