Learning Plasma Dynamics and Robust Rampdown Trajectories with Predict-First Experiments at TCV

2025-02-17

AI Summary

Modeling plasma current rampdown in tokamaks is challenging due to susceptibility to thermal instabilities, edge-localized mode (ELM) crashes, and other operational constraints, leading to high disruption risk. This work proposes a physics-informed neural state-space model (NSSM) integrated with proximal policy optimization (PPO)-based deep reinforcement learning to generate robust, constraint-aware rampdown trajectories, introducing a novel “prediction-first” experimental paradigm. Leveraging only five high-performance discharges, the approach achieves few-shot generalization and, for the first time on TCV, demonstrates NSSM+RL’s capability to simultaneously avoid multiple instability boundaries. Experimental results show significantly improved termination accuracy for plasma current and normalized energy; reliable discharges with 20% higher current (100% successful termination); and trajectory robustness exceeding 95%.

Technology Category

• Intelligent Robots: State Estimation
• Machine Learning: Reinforcement Learning
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

The rampdown in tokamak operations is a difficult to simulate phase during which the plasma is often pushed towards multiple instability limits. To address this challenge, and reduce the risk of disrupting operations, we leverage recent advances in Scientific Machine Learning (SciML) to develop a neural state-space model (NSSM) that predicts plasma dynamics during Tokamak `a Configuration Variable (TCV) rampdowns. By integrating simple physics structure and data-driven models, the NSSM efficiently learns plasma dynamics during the rampdown from a modest dataset of 311 pulses with only five pulses in the reactor relevant high performance regime. The NSSM is parallelized across uncertainties, and reinforcement learning (RL) is applied to design trajectories that avoid multiple instability limits with high probability. Experiments at TCV ramping down high performance plasmas show statistically significant improvements in current and energy at plasma termination, with improvements in speed through continuous re-training. A predict-first experiment, increasing plasma current by 20% from baseline, demonstrates the NSSM's ability to make small extrapolations with sufficient accuracy to design trajectories that successfully terminate the pulse. The developed approach paves the way for designing tokamak controls with robustness to considerable uncertainty, and demonstrates the relevance of the SciML approach to learning plasma dynamics for rapidly developing robust trajectories and controls during the incremental campaigns of upcoming burning plasma tokamaks.

Problem

• Predicting plasma dynamics during tokamak rampdowns.
• Designing trajectories to avoid plasma instabilities.
• Improving plasma termination with machine learning models.

Innovation

• Neural state-space model predicts plasma dynamics.
• Integration of physics structure and data-driven models.
• Reinforcement learning designs instability-avoiding trajectories.