Automatic detection of L-H transition in KSTAR by support vector machine

Shin, Giwook; Juhn, June-Woo; Kwon, Giil; Son, Seok-Kwon; Hahn, S.H.

doi:10.1016/j.fusengdes.2017.12.011

Cited by 6 publications

(6 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The patterns of the first L-H transition and ELMs in KSTAR fusion plasmas can be classified by the diagnostic features of D α emission and line-averaged electron density [18,19] because the time-series diagnostic signals have characteristic patterns that allow the classification of the H-mode transition and the ELMs. As previously reported in reference [19], the LSTM, which is a supervised machine learning method, performs well for memorizing patterns from long sequential data with a relatively short inference time and high classification accuracy.…”

Section: Description Of Lstm Classifier In Kstar Pcsmentioning

confidence: 99%

“…The onset of the RMP-driven ELM control can be automated if we can determine in real-time when the first L-H transition occurs. A machine-learning (ML) approach has advantages for such automation of preemptive control [18,19]. Based on the output of an ML classifier that classifies the plasma states as L-mode, H-mode, and ELMy, a control algorithm has been implemented in the KSTAR plasma control system (PCS) [20] that triggers the use of RMP based on the real-time change in the ML classifier output, which indicates the first L-H transition and the first ELM burst.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Preemptive RMP-driven ELM crash suppression automated by a real-time machine-learning classifier in KSTAR

Shin¹,

Hahn²,

Kim³

et al. 2022

Nucl. Fusion

Self Cite

View full text Add to dashboard Cite

Suppression or mitigation of edge-localized mode (ELM) crashes is necessary for ITER. The strategy to suppress all the ELM crashes by the resonant magnetic perturbation (RMP) should be applied as soon as the first low-to-high confinement (L-H) transition occurs. A control algorithm based on real-time machine learning (ML) enables such an approach: it classifies the H-mode transition and the ELMy phase in real-time and automatically applies the preemptive RMP. This paper reports the algorithm design, which is now implemented in the KSTAR plasma-control system, and the corresponding experimental demonstration of typical high-δ KSTAR H-mode plasmas. As a result, all initial ELM crashes are suppressed with an acceptable safety factor at the edge (q95) and with RMP field adjustment. Moreover, the ML-driven ELM-crash-suppression discharges remain stable without further degradation due to the regularization of the plasma pedestal.

show abstract

Section: Description Of Lstm Classifier In Kstar Pcsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Preemptive RMP-driven ELM crash suppression automated by a real-time machine-learning classifier in KSTAR

Shin¹,

Hahn²,

Kim³

et al. 2022

Nucl. Fusion

Self Cite

View full text Add to dashboard Cite

show abstract

“…To this end, several machine learning (ML) [29] methods have been developed in recent years in order to work toward the common goal of real-time plasma confinement regime classification. Specifically, a flurry of both classical [30][31][32] and deep [33][34][35][36][37][38][39] learning approaches have been pursued to classify confinement regimes on Alcator C-Mod [31], KSTAR [32,34], TCV [35,36], COMPASS [37], DIII-D [38], and EAST [39]. These previous approaches exhibit varying degrees of real-time implementation, with some [35,36] primarily used in an offline setting, and others [38] successfully ran in a real-time control scheme.…”

Section: Introductionmentioning

confidence: 99%

Real-time confinement regime detection in fusion plasmas with convolutional neural networks and high-bandwidth edge fluctuation measurements

Gill,

Smith,

Joung

et al. 2024

Mach. Learn.: Sci. Technol.

View full text Add to dashboard Cite

A real-time detection of the plasma confinement regime can enable new advanced plasma control capabilities for both the access to and sustainment of enhanced confinement regimes in fusion devices. For example, a real-time indication of the confinement regime can facilitate transition to the high-performing wide pedestal quiescent H-mode, or avoid unwanted transitions to lower confinement regimes that may induce plasma termination. To demonstrate real-time confinement regime detection, we use the 2D beam emission spectroscopy (BES) diagnostic system to capture localized density fluctuations of long wavelength turbulent modes in the edge region at a 1 MHz sampling rate. BES data from 330 discharges in either L-mode, H-mode, Quiescent H (QH)-mode, or wide-pedestal QH-mode was collected from the DIII-D tokamak and curated to develop a high-quality database to train a deep-learning classification model for real-time confinement detection. We utilize the 6x8 spatial configuration with a time window of 1024 $\mu$s and recast the input to obtain spectral-like features via FFT preprocessing. We employ a shallow 3D convolutional neural network for the multivariate time-series classification task and utilize a softmax in the final dense layer to retrieve a probability distribution over the different confinement regimes. Our model classifies the global confinement state on 44 unseen test discharges with an average $F_1$ score of 0.94, using only $\sim$1 millisecond snippets of BES data at a time. This activity demonstrates the feasibility for real-time data analysis of fluctuation diagnostics in future devices such as ITER, where the need for reliable and advanced plasma control is urgent.

show abstract

“…is amount of data implies performing the analysis (e.g., finding significance or regular patterns) in high dimensional spaces and it is essential to automate the process using machine learning [6][7][8][9]. To this end, we can find several algorithms in the literature in order to perform pattern recognition in an automatic way.…”

Section: Introductionmentioning

confidence: 99%

Nuclear Fusion Pattern Recognition by Ensemble Learning

et al. 2021

View full text Add to dashboard Cite

Nuclear fusion is the process by which two or more atomic nuclei join together to form a single heavier nucleus. This is usually accompanied by the release of large quantities of energy. This energy could be cheaper, cleaner, and safer than other technology currently in use. Experiments in nuclear fusion generate a large number of signals that are stored in huge databases. It is impossible to do a complete analysis of this data manually, and it is essential to automate this process. That is why machine learning models have been used to this end in previous years. In the literature, several popular algorithms can be found to carry out the automatic classification of signals. Among these, ensemble methods provide a good balance between success rate and internal information about models. Specifically, AdaBoost algorithm will allow obtaining an explicit set of rules that explains the class for each input data, adding interpretability to the models. In this paper, an innovative approach to perform an online classification, that is, to identify the discharge before it actually ends, using interpretable models is presented. In order to evaluate and reveal the benefits of rule-based models, an illustrative example has been implemented to perform an online classification of five different signals of the TJ-II stellarator fusion device located in Madrid, Spain.

show abstract

Automatic detection of L-H transition in KSTAR by support vector machine

Cited by 6 publications

References 9 publications

Preemptive RMP-driven ELM crash suppression automated by a real-time machine-learning classifier in KSTAR

Preemptive RMP-driven ELM crash suppression automated by a real-time machine-learning classifier in KSTAR

Real-time confinement regime detection in fusion plasmas with convolutional neural networks and high-bandwidth edge fluctuation measurements

Nuclear Fusion Pattern Recognition by Ensemble Learning

Contact Info

Product

Resources

About