The application of classification methods in a data driven investigation of the JET L–H transition

During a tokamak discharge, the plasma can vary between different confinement regimes: Low (L), High (H) and, in some cases, a temporary (intermediate state), called Dithering (D). In addition, while the plasma is in H mode, Edge Localized Modes (ELMs) can occur. The automatic detection of changes between these states, and of ELMs, is important for tokamak operation. Motivated by this, and by recent developments in Deep Learning (DL), we developed and compared two methods for automatic detection of the occurrence of L-D-H transitions and ELMs, applied on data from the TCV tokamak. These methods consist in a Convolutional Neural Network (CNN) and a Convolutional Long Short Term Memory Neural Network (Conv-LSTM). We measured our results with regards to ELMs using ROC curves and Youden's score index, and regarding state detection using Cohen's Kappa Index.

show abstract

“…Neural Networks [28] on data from several tokamaks for detection of L-H transitions, classification of L and H modes, and detection of ELMs.…”

Section: Previous Workmentioning

confidence: 99%

Classification of tokamak plasma confinement states with convolutional recurrent neural networks

et al. 2020

View full text Add to dashboard Cite

show abstract

“…The database, used to assess the performance of the predictor developed for JET, is a set of 50 discharges, with the divertor MARKII gas box with the septum, for which the transition times have been determined by the experts with a high degree of confidence [19]. Only for 42 shots all signals are available during 2 s segments around the transition from low (L) to high (H) confinement regime.…”

Section: Confinement Regime Identificationmentioning

confidence: 99%

Innovative signal processing and data analysis methods on JET for control in the perspective of next-step devices

Murari¹,

Vega²,

Mazon³

et al. 2010

Nucl. Fusion

View full text Add to dashboard Cite

In the last few years, it has been realized that more sophisticated control schemes are necessary to push the boundaries of tokamak operation and the performance of reactor-like machines. In addition, JET needs to operate safely with the new metallic wall and such protection will be needed for ITER. These objectives have motivated the development, benchmark and validation of new signal processing and data analysis methods. Two new approaches for the determination of the magnetic topology in real time have been validated on an extensive database of JET discharges, including advanced tokamak scenarios. Robust methods of confinement regime identification and disruption prediction are a prerequisite for safe, general control schemes. New identifiers have been developed and their success rates exceed 99% in determining whether plasmas are in the L or H mode. A new disruption predictor is being developed and has already provided success rates higher than 90% in realistic real-time conditions. Moreover, the generalization capability of this new predictor has been confirmed by applying it to new experimental campaigns not used for the training. The success rate remains high even more than ten campaigns, or about four years, after the last one used for the training. The deployment of video cameras in real time requires the development of new image processing algorithms, which have already been implemented and validated successfully on JET for the real-time identification of hot spots with a time resolution of tens of milliseconds. A series of new feedback schemes has also been explicitly developed not much to control the plasma but to really improve the physics understanding of some phenomena. Particularly interesting are the simultaneous control of the safety factor and pressure profiles and the real-time tracking of toroidal Alfven eigenmode instabilities. These advanced feedback schemes for physics understanding often require more advanced signal processing techniques like adaptive filtering, which have already been implemented. The paper concludes by discussing the use of these real-time analysis and control developments in next-step machines such as ITER.

show abstract

“…For that reason, work has been put into developing tools capable of automating the task of detecting different confinement modes. In particular, in the past few years, research has been done on using Machine Learning [6,7,8,9,10] and, more recently, Deep Learning [1] for this task. These algorithms are particularly suitable for dealing with such challenges of extracting patterns of such high-dimensional data collected during these experiments.…”

Section: Introductionmentioning

confidence: 99%

Plasma Confinement Mode Classification Using a Sequence-to-Sequence Neural Network With Attention

Matos,

Menkovski,

Pau

et al. 2020

Preprint

View full text Add to dashboard Cite

In a typical fusion experiment, the plasma can have several possible confinement modes. At the TCV tokamak, aside from the Low (L) and High (H) confinement modes, an additional mode, dithering (D), is frequently observed. Developing methods that automatically detect these modes is considered to be important for future tokamak operation. Previous work [1] with deep learning methods, particularly convolutional recurrent neural networks (Conv-RNNs), indicates that they are a suitable approach. Nevertheless, those models are sensitive to noise in the temporal alignment of labels, and that model in particular is limited to making individual decisions taking into account only its own hidden state and its input at each time step. In this work, we propose an architecture for a sequence-to-sequence neural network model with attention which solves both of those issues. Using a carefully calibrated dataset, we compare the performance of a Conv-RNN with that of our proposed sequenceto-sequence model, and show two results: one, that the Conv-RNN can be improved upon with new data; two, that the sequence-to-sequence model can improve the results even further, achieving excellent scores on both train and test data.

show abstract

The application of classification methods in a data driven investigation of the JET L–H transition

Cited by 13 publications

References 32 publications

Classification of tokamak plasma confinement states with convolutional recurrent neural networks

Classification of tokamak plasma confinement states with convolutional recurrent neural networks

Innovative signal processing and data analysis methods on JET for control in the perspective of next-step devices

Plasma Confinement Mode Classification Using a Sequence-to-Sequence Neural Network With Attention

Contact Info

Product

Resources

About