Feature selection for disruption prediction from scratch in JET by using genetic algorithms and probabilistic predictors

Pereira, A.; Vega, J.; Moreno, R.; Dormido-Canto, S.; Rattá, G.A.; Pavón, Fernando

doi:10.1016/j.fusengdes.2015.04.040

Cited by 3 publications

(2 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The requirement of a large database of disruptive discharges to train the predictor is not compatible with safe ITER and DEMO operations. Therefore, adaptive predictors have been proposed with high learning rates using a limited disruption database [101][102][103][104]. The strategy is to retrain the predictors on a wider database only after missed disruption detection.…”

Section: Disruption Predictionmentioning

confidence: 99%

Overview of the JET results in support to ITER

Litaudon¹,

Abduallev²,

Abhangi³

et al. 2017

Nucl. Fusion

161

View full text Add to dashboard Cite

The 2014–2016 JET results are reviewed in the light of their significance for optimising the ITER research plan for the active and non-active operation. More than 60 h of plasma operation with ITER first wall materials successfully took place since its installation in 2011. New multi-machine scaling of the type I-ELM divertor energy flux density to ITER is supported by first principle modelling. ITER relevant disruption experiments and first principle modelling are reported with a set of three disruption mitigation valves mimicking the ITER setup. Insights of the L–H power threshold in Deuterium and Hydrogen are given, stressing the importance of the magnetic configurations and the recent measurements of fine-scale structures in the edge radial electric. Dimensionless scans of the core and pedestal confinement provide new information to elucidate the importance of the first wall material on the fusion performance. H-mode plasmas at ITER triangularity (H = 1 at βN ~ 1.8 and n/nGW ~ 0.6) have been sustained at 2 MA during 5 s. The ITER neutronics codes have been validated on high performance experiments. Prospects for the coming D–T campaign and 14 MeV neutron calibration strategy are reviewed.

show abstract

Section: Disruption Predictionmentioning

confidence: 99%

Overview of the JET results in support to ITER

Litaudon¹,

Abduallev²,

Abhangi³

et al. 2017

Nucl. Fusion

161

View full text Add to dashboard Cite

show abstract

“…The AUC value [36], which is the area under the receiver operation characteristic (ROC) curve [37], can be used to evaluate the quality of different models. In a trained model, each threshold corresponds to a set of pairs (S , p S n ).…”

Section: Density Limit Prediction On East Using Lstm With Different M...mentioning

confidence: 99%

Density limit disruption prediction using a long short-term memory network on EAST

Zhang

Chen

Guo

et al. 2020

Plasma Sci. Technol.

View full text Add to dashboard Cite

Disruption prediction using a long short-term memory (LSTM) algorithm has been developed on EAST, due to its inherent advantages in time series data processing. In the present work, LSTM is used as the model and the AUC (area under receiver operation characteristic curve) is used as the evaluation index. When the model is trained on data from the plasma current flattop phase and tested on data from the same period multiple times, the highest AUC is 0.8646 and the training time is about 6900 s per epoch. For comparison, the last 1000 ms of the flattop phases are intercepted as short time sequences. When the model is trained on data from short time sequences and tested on data from the same period, the highest AUC is increased to 0.9379 and the training time is restricted to 36 s per epoch. When the best model trained on the short time sequences is applied to the flattop phase for testing, the AUC is up to 0.9189. The experiment results show that it is possible for LSTM to train the model on data from short time sequences and migrate the model to the entire flattop phase, with a shorter training time and higher AUC value.

show abstract

Classification of Muscle Fatigue in Dynamic Contraction Using Surface Electromyography Signals and Multifractal Singularity Spectral Analysis

Marri

Swaminathan

2016

Journal of Dynamic Systems, Measurement, and Control

View full text Add to dashboard Cite

Muscle fatigue is a neuromuscular condition experienced during daily activities. This phenomenon is generally characterized using surface electromyography (sEMG) signals and has gained a lot of interest in the fields of clinical rehabilitation, prosthetics control, and sports medicine. sEMG signals are complex, nonstationary and also exhibit self-similarity fractal characteristics. In this work, an attempt has been made to differentiate sEMG signals in nonfatigue and fatigue conditions during dynamic contraction using multifractal analysis. sEMG signals are recorded from biceps brachii muscles of 42 healthy adult volunteers while performing curl exercise. The signals are preprocessed and segmented into nonfatigue and fatigue conditions using the first and last curls, respectively. The multifractal detrended moving average algorithm (MFDMA) is applied to both segments, and multifractal singularity spectrum (SSM) function is derived. Five conventional features are extracted from the singularity spectrum. Twenty-five new features are proposed for analyzing muscle fatigue from the multifractal spectrum. These proposed features are adopted from analysis of sEMG signals and muscle fatigue studies performed in time and frequency domain. These proposed 25 feature sets are compared with conventional five features using feature selection methods such as Wilcoxon rank sum, information gain (IG) and genetic algorithm (GA) techniques. Two classification algorithms, namely, k-nearest neighbor (k-NN) and logistic regression (LR), are explored for differentiating muscle fatigue. The results show that about 60% of the proposed features are statistically highly significant and suitable for muscle fatigue analysis. The results also show that eight proposed features ranked among the top 10 features. The classification accuracy with conventional features in dynamic contraction is 75%. This accuracy improved to 88% with k-NN-GA combination with proposed new feature set. Based on the results, it appears that the multifractal spectrum analysis with new singularity features can be used for clinical evaluation in varied neuromuscular conditions, and the proposed features can also be useful in analyzing other physiological time series.

show abstract

Feature selection for disruption prediction from scratch in JET by using genetic algorithms and probabilistic predictors

Cited by 3 publications

References 7 publications

Overview of the JET results in support to ITER

Overview of the JET results in support to ITER

Density limit disruption prediction using a long short-term memory network on EAST

Classification of Muscle Fatigue in Dynamic Contraction Using Surface Electromyography Signals and Multifractal Singularity Spectral Analysis

Contact Info

Product

Resources

About