Deep convolutional neural networks for multi-scale time-series classification and application to tokamak disruption prediction using raw, high temporal resolution diagnostic data

Churchill, R.M.; Tobias, Benjamin; Zhu, Yilun; Team, Diii-D

doi:10.1063/1.5144458

Cited by 57 publications

(38 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Compared with the previously published models based on the LSTM architecture, FRNN models constructed with the temporal convolutional neural network (TCN) architecture exhibit at least equivalent and demonstrably superior computational performance and predictive power for disruption forecasting across various experimental databases from the DIII-D and JET tokamaks. The TCN architecture has also been applied in other disruption studies recently, using input from the Electron Cyclotron Emission imaging (ECEi) diagnostic data on DIII-D (Churchill et al, 2020).…”

Section: Summary and Future Workmentioning

confidence: 99%

Fully Convolutional Spatio-Temporal Models for Representation Learning in Plasma Science

Dong

Felker

Svyatkovskiy

et al. 2021

J Mach Learn Model Comput

View full text Add to dashboard Cite

We have trained a fully convolutional spatio-temporal model for fast and accurate representation learning in the challenging exemplar application area of fusion energy plasma science. The onset of major disruptions is a critically important fusion energy science issue that must be resolved for advanced tokamak plasmas such as the $25B burning plasma international thermonuclear experimental reactor (ITER) experiment. While a variety of statistical methods have been used to address the problem of tokamak disruption prediction and control, recent approaches based on deep learning have proven particularly compelling. In the present paper, we introduce further improvements to the fusion recurrent neural network (FRNN) software suite, which delivered cross-machine disruption predictions with unprecedented accuracy using a large database of experimental signals from two major tokamaks. Up to now, FRNN was based on the long short-term memory (LSTM) variant of recurrent neural networks to leverage the temporal information in the data. Here, we implement and apply the "temporal convolutional neural network (TCN)" architecture to the time-dependent input signals. This allows highly optimized convolution operations to carry the majority of the computational load of training, thus enabling a reduction in training time, and the effective use of highperformance computing resources for hyperparameter tuning. At the same time, the TCN-based architecture achieves better predictive performance when compared with the LSTM architecture for various tasks for a representative fusion database.

show abstract

Section: Summary and Future Workmentioning

confidence: 99%

Fully Convolutional Spatio-Temporal Models for Representation Learning in Plasma Science

Dong

Felker

Svyatkovskiy

et al. 2021

J Mach Learn Model Comput

View full text Add to dashboard Cite

show abstract

“…III. These include further work into neural network architectures for multiscale fusion diagnostic data 28 with the ability to combine multiple diagnostics in predictions. An important ML tool to develop is a model to extract physical model parameters from diagnostic data.…”

Section: Ongoing and Future Workmentioning

confidence: 99%

“…27 An example of the ML algorithms/techniques to be incorporated are the temporal convolution networks (TCNs) for detection of events in diagnostic time-series data, with a recent example applying TCN to detect disruptions using electron cyclotron emission imaging (ECEI) data on DIII-D (Ref. 28). Additionally, the framework will be worked to connect to the various analysis codes available in the Integrated Modeling and Analysis Suite 29 (IMAS), including potential future work on IDA, and workflows defined in the One Modeling Framework for Integrated Tasks (OMFIT) framework.…”

mentioning

confidence: 99%

A Framework for International Collaboration on ITER Using Large-Scale Data Transfer to Enable Near-Real-Time Analysis

Churchill

Chang

Choi

et al. 2021

Fusion Science and Technology

Self Cite

View full text Add to dashboard Cite

The global nature of the ITER project along with its projected approximately petabyte-per-day data generation presents not only a unique challenge but also an opportunity for the fusion community to rethink, optimize, and enhance our scientific discovery process. Recognizing this, collaborative research with computational scientists was undertaken over the past several years to create a framework for large-scale data movement across wide-area networks to enable global near-real-time analysis of fusion data. This would broaden the available computational resources for analysis/simulation and increase the number of researchers actively participating in experiments.An official demonstration of this framework for fast, large data transfer and real-time analysis was carried out between the KSTAR tokamak in Daejeon, Korea, and Princeton Plasma Physics Laboratory (PPPL) in Princeton, New Jersey. Streaming large data transfer, with near-real-time movie creation and analysis of the KSTAR electron cyclotron emission imaging data, was performed using the Adaptable Input Output (I/O) System (ADIOS) framework, and comparisons were made at PPPL with simulation results from the XGC1 code. These demonstrations were made possible utilizing an optimized network configuration at PPPL, which achieved over 8.8 Gbps (88% utilization) in throughput tests from the National Fusion Research Institute to PPPL. This demonstration showed the feasibility for large-scale data analysis of KSTAR data and provides a nascent framework to enable use of globally distributed computational and personnel resources in pursuit of scientific knowledge from the ITER experiment.

show abstract

“…High dimensional data is a feedforward learning data processing model, and its structure is very similar to radial basis function. Compared with the RBF data processing model, the generalized high-dimensional data has more advantages in approximation ability and convergence speed [5][6] . The multivariable time series pre prediction method uses the correlation between multiple time series to improve the overall prediction accuracy.…”

Section: Multivariate Time Series Prediction Of High Dimensional Data 21multivariate Time Series Association Algorithm For High Dimensionmentioning

confidence: 99%

“…If each variable chooses the appropriate time delay τ and embedding dimension m i (i = 1, 2…n). Among them, τi and mi are the key to the success or failure of phase space reconstruction [7][8] . Through the above formula, the regression estimation is obtained, and the predicted time series are obtained.…”

Section: Multivariate Time Series Prediction Of High Dimensional Data 21multivariate Time Series Association Algorithm For High Dimensionmentioning

confidence: 99%

Multivariate time series prediction of high dimensional data based on deep reinforcement learning

Zhang

et al. 2021

E3S Web Conf.

View full text Add to dashboard Cite

In order to improve the prediction accuracy of high-dimensional data time series, a high-dimensional data multivariate time series prediction method based on deep reinforcement learning is proposed. The deep reinforcement learning method is used to solve the time delay of each variable and mine the data characteristics. According to the principle of maximum conditional entropy, the embedding dimension of the phase space is expanded, and a multivariate time series model of high-dimensional data is constructed. Thus, the conversion of reconstructed coordinates from low-dimensional to high-dimensional can be kept relatively stable. The strong independence and low redundancy of the final reconstructed phase space construct an effective model input vector for multivariate time series forecasting. Numerical experiments of classical multivariable chaotic time series show that the method proposed in this paper has better forecasting effect, which shows the forecasting effectiveness of this method.

show abstract

Deep convolutional neural networks for multi-scale time-series classification and application to tokamak disruption prediction using raw, high temporal resolution diagnostic data

Cited by 57 publications

References 39 publications

Fully Convolutional Spatio-Temporal Models for Representation Learning in Plasma Science

Fully Convolutional Spatio-Temporal Models for Representation Learning in Plasma Science

A Framework for International Collaboration on ITER Using Large-Scale Data Transfer to Enable Near-Real-Time Analysis

Multivariate time series prediction of high dimensional data based on deep reinforcement learning

Contact Info

Product

Resources

About