physics.plasm-phAug 23, 2026

Interpretable statistical feature engineering for early disruption prediction in the short pulse ADITYA tokamak

Authors: Jyoti Agarwal, Kavit Patel, Bhaskar Chaudhury, Abhishek Sharma, Shrichand Jakhar, Manika Sharma

Organizations: Group in Computational Science and HPC, Dhirubhai Ambani University (formerly DA-IICT), Gandhinagar, INDIA. · Institute for Plasma Researach, Gandhinagar, India - 382428

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Reliable early disruption prediction is critical for the safe operation and real-time control of tokamaks. However, machine learning based prediction frameworks have predominantly targeted medium and long pulse devices, with comparatively limited attention given to short pulse tokamaks where available warning time is inherently constrained. In this work, an interpretable machine learning framework is developed for feature engineering and early prediction of disruptions in the ADITYA using the initial plasma evolution information, prior to the activation of the negative converter of the ohmic transformer power supply. Statistical descriptors comprising the mean, variance, skewness, kurtosis and wavelet energy entropy are extracted from routinely available plasma diagnostics over different operation time windows. Decision tree based feature selection is employed to identify physically meaningful disruption precursors and to reduce feature dimensionality. These selected features are used to train a random forest classifier. The proposed framework achieves stable predictive performance across different analysis windows, with a maximum ROC-AUC of 0.87 for 0-35 ms and 0-40 ms windows. Comparable and in some cases improved, performance is obtained using the reduced feature set, demonstrating that the selected statistical descriptors retain the essential information required for disruption prediction. The proposed methodology provides an interpretable and computationally efficient framework for real time disruption prediction in short pulse tokamaks and establishes that carefully engineered statistical descriptors can effectively replace raw time series inputs for early disruption prediction, thereby offering a practical pathway toward real time plasma control in short pulse tokamaks similar to ADITYA and ADITYA-U.

Figures & tables

Figure 1: Schematic overview of the proposed interpretable ML framework for early disruption prediction in the ADITYA tokamak
Statistical descriptorNomenclature
Variance in plasma currentIpvIp_{v}
Mean of plasma currentIpmIp_{m}
Skewness of plasma currentIpsIp_{s}
Kurtosis of plasma currentIpkIp_{k}
Wavelet energy entropy of plasma currentIpweeIp_{wee}
Variance in bolometerBolovBolo_{v}
Table 1: Statistical descriptor passed as features for specific observation window
Figure 2: Flowchart of the CART decision tree classifier
ParameterValue
n_estimators200
max_depthNone
max_featuressqrt
min_samples_split2
min_samples_leaf1
Table 2: Random forest hyperparameters used in the present study.
Time windowImportant feature
0-20 msBolos, Bolov, C111k, C111m, C111s, Hαm, Hαwee, HXRm, HXRv, Ipm, SXRm, SXRvBolo_{s},\,Bolo_{v},\,C111_{k},\,C111_{m},\,C111_{s},\,H\alpha_{m},\,H\alpha_{wee},\,HXR_{m},\,HXR_{v},\,Ip_{m},\,SXR_{m},\,SXR_{v}
0-25 msBolom, Bolov, C111m, HXRm, HXRv, Ipm, Ipwee, SXRmBolo_{m},\,Bolo_{v},\,C111_{m},\,HXR_{m},\,HXR_{v},\,Ip_{m},\,Ip_{wee},\,SXR_{m}
0-30 msBolos, Bolov, C111k, C111m, Hαv, HXRk, HXRwee, HXRm, Ipk, Ipv, SXRm, SXRvBolo_{s},\,Bolo_{v},\,C111_{k},\,C111_{m},\,H\alpha_{v},\,HXR_{k},\,HXR_{wee},\,HXR_{m},\,Ip_{k},\,Ip_{v},\,SXR_{m},\,SXR_{v}
0-35 msBolom, Bolowee, C111m, HXRk, HXRwee, Ipm, Ipv, Ipwee, SXRmBolo_{m},\,Bolo_{wee},\,C111_{m},\,HXR_{k},\,HXR_{wee},\,Ip_{m},\,Ip_{v},\,Ip_{wee},\,SXR_{m}
0-40 msBolom, Bolov,  Bolowee, C111m, HXRwee, HXRv, Ipm, Ipv, Ipwee, SXRm, SXRweeBolo_{m},\,Bolo_{v},\,\,Bolo_{wee},\,C111_{m},\,HXR_{wee},\,HXR_{v},\,Ip_{m},\,Ip_{v},\,Ip_{wee},\,SXR_{m},\,SXR_{wee}
0-45 msC111m, C111s, Hαs, HXRv, HXRwee, Ipm, Ipv, Ipwee, SXRweeC111_{m},\,C111_{s},\,H\alpha_{s},\,HXR_{v},\,HXR_{wee},\,Ip_{m},\,Ip_{v},\,Ip_{wee},\,SXR_{wee}
Table 3: Important parameters with respect to experimental time window
Time windowRoot node feature selected from each diagnostic
BolometerCC -IIIHαH\alphaHXRIpI_{p}SXR
0–20 msBolovBolo_{v}CIIImCIII_{m}HαweeH\alpha_{wee}HXRsHXR_{s}Ip,sI_{p,s}SXRmSXR_{m}
0–25 msBolovBolo_{v}CIIImCIII_{m}HαvH\alpha_{v}HXRvHXR_{v}Ip,vI_{p,v}SXRmSXR_{m}
0–30 msBolovBolo_{v}CIIImCIII_{m}HαvH\alpha_{v}HXRvHXR_{v}Ip,vI_{p,v}SXRmSXR_{m}
0–35 msBolosBolo_{s}CIIImCIII_{m}HαvH\alpha_{v}HXRvHXR_{v}Ip,vI_{p,v}SXRmSXR_{m}
0–40 msBolosBolo_{s}CIIImCIII_{m}HαvH\alpha_{v}HXRvHXR_{v}Ip,vI_{p,v}SXRmSXR_{m}
Table 4: Root node feature selected by the DT classifier for each diagnostic signal across different analysis windows.
Figure 3: ROC curves for different time windows
Figure 4: Variation of the metrics (a) accuracy, (b) ROC-AUC score, (c) precision and (d) recall for RF classifier using complete feature set and reduced feature set with sample time window
Complete feature model
 Pred. Non DisruptivePred. Disruptive
Actual Non Disruptive10325
Actual Disruptive2979
Reduced feature model
 Pred. Non DisruptivePred. Disruptive
Actual Non Disruptive10523
Table 5: Confusion matrices for the 40 ms prediction window.

Explore similar work

CardsList
Sep 21, 2026

Horizon-Aware Early Event Prediction for Tokamak Disruption Alarms

Disruption predictors must warn the plasma control system early enough for the intended response, and the required warning time differs between mitigation and avoidance. We view a warning system as a predictor that outputs a disruption risk and an alarm policy that acts on it, and ask which risk the predictor should learn. Binary classifiers do not specify when a disruption occurs, and survival analysis models the full distribution of the time to disruption, whereas an alarm depends only on the risk within a finite horizon. We introduce two horizon-focused targets into tokamak disruption prediction: a horizon-truncated survival objective and fixed-horizon early event prediction (EEP) with temporal label smoothing (TLS). We compare them with binary classification and survival analysis on the DIII-D, Alcator C-Mod, and EAST databases of DisruptionBench, evaluating the risk by its threshold-free ranking at a 40 ms deadline and the alarm policy by thresholds calibrated on validation data. On every device, TLS outperforms the deep survival models in both ranking and alarm performance; TLS performs best on DIII-D and EAST, and a random forest trained on a hard horizon target performs best on C-Mod. TLS also improves the calibrated alarm performance over hard targets placed at either end of its smoothing range on all devices, isolating the benefit of label smoothing. Horizon truncation alone did not consistently improve survival analysis. Alarm performance remains limited on C-Mod and DIII-D and varies strongly across training seeds, and ablations show that the target should extend to the required warning time.

Takeshi Koshizuka, Takaharu Yaguchi

RIKEN AIP, Japan. · Kyushu University IMI, Fukuoka, Japan.

Event Time PredictionSurvival Analysis
AskPDF
Jul 5, 2026

Hierarchical Multi-to-Single-Modal Knowledge Distillation for Disruption Prediction in EAST

Plasma disruption is a critical threat to tokamak safety. Existing data-driven predictors mainly rely on time-series diagnostic signals, while visible images provide complementary spatial cues including plasma deformation, local brightening, and radiation-structure evolution. Although the image modality improves the model's discriminative capability, it also substantially increases the computational cost during inference. To address this issue, we propose a hierarchical multi-to-single-modal knowledge distillation framework for disruption prediction on a synchronized EAST multimodal dataset. During training, visible images and time-series signals are used to train a multimodal teacher, which learns disruption precursor representations through Transformer-based encoders and a prototype-guided spatiotemporal hypergraph module. During inference, only the time-series student is retained, with multimodal knowledge transferred through graph-structure-level, representation-level, and decision-level distillation. On the 640-discharge EAST dataset, the results demonstrate that the proposed framework can preserve the discriminative advantages of multimodal learning while substantially reducing inference cost, and providing an effective route for efficient disruption prediction in EAST. The source code of this paper will be released on https://github.com/Event-AHU/OpenFusion.

Qiang Chen, Xiao Wang, Hao Si +9

School of Computer Science and Technology, Anhui University, Hefei 230031, China · Institute of Plasma Physics, Chinese Academy of Sciences, Hefei 230031, China · State Key Laboratory of Opto-Electronic Information Acquisition and Protection Technology, Anhui University, Hefei 230039, China

Cross-Modal Knowledge DistillationEfficient Multimodal InferenceMultimodal Sensor Fusion
AskPDF
Jul 5, 2026

Benchmarking Sensor Robustness in Plasma Diagnostic Models: A Systematic Evaluation on TokaMark

Plasma diagnostic models for tokamak fusion devices are almost universally evaluated on clean, complete sensor data. In practice, fusion diagnostics fail regularly: acquisition systems start late, individual sensors die, and signal dropouts cluster precisely when a plasma disruption is approaching. We present the first systematic robustness benchmark for plasma diagnostic ML using the TokaMark dataset of 11,573 MAST shots, evaluating XGBoost, LSTM, Transformer, and the TokaMark CNN baseline across six physically-grounded failure scenarios and three imputation strategies. We introduce the Robustness Score (RS) for standardized cross-architecture comparison. Our central finding is that disruption-proximate sensor failure (corruption injected in the final window timesteps) collapses sequence model performance (LSTM +212% NRMSE) while a statistical feature model remains comparatively stable (XGBoost +37%). Forward-fill imputation eliminates nearly all degradation from random dropout for sequence models (LSTM +57% to ~0%), but offers little help when the end of the window is corrupted. Shot-level alarm evaluation using ground-truth disruption timestamps reveals that LSTM alarm detection collapses to TPR=0.00 under proximate sensor failure, while mean-fill imputation recovers it to TPR=1.00, a reversal of the pattern observed in NRMSE. Plasma current emerges as the single most critical diagnostic across all architectures (+73% to +140% upon removal). Code, data, and trained checkpoints are available at https://github.com/Neerav-Gupta/tokamark-robustness.

Neerav Gupta

Independent Researcher

Incomplete Data Imputation
AskPDF