A data-driven model for stable long-horizon autoregressive prediction of plasma current and control-oriented boundary evolution in EAST
Open Access
- 28 September 2026
- journal article
- Published by IOP Publishing in Nuclear Fusion
- Vol. 66 (12) , 124004
Abstract
Accurate long-horizon prediction of tokamak plasma current, position, and boundary evolution is essential for magnetic control, rapid controller development, and reinforcement-learning-based optimization. However, high-fidelity physics-based simulators are often computationally prohibitive when large numbers of simulations are required for controller tuning and large-scale optimization. In this work, we develop a fast, data-driven model for the Experimental Advanced Superconducting Tokamak (EAST) to predict plasma current, position, and shape evolution over horizons of up to 1 s, with an inference time of 0.1 s on a single NVIDIA H800 GPU. Evaluated on a temporally separated test set comprising 5907 discharges, the model demonstrates strong generalization to evolving operating conditions and maintains stable agreement with experimental measurements during long-horizon autoregressive rollouts. The mean geometric error of the plasma centroid and the control-oriented boundary representation is 2.4 cm, indicating reliable centimeter-level accuracy for control-oriented prediction of plasma position and shape evolution. The proposed framework provides a practical, high-throughput alternative to first-principles simulators, enabling efficient control algorithm prototyping, large-scale scenario exploration, and future data-driven optimization for tokamak plasma control. This study focuses on the post-shaping phase of EAST discharges, where sustained plasma current and control-oriented boundary regulation are most critical.Keywords
Funding Information
- Science Foundation of Institute of Plasma Physics, Chinese Academy of Sciences (DSJJ-2025-04)
- Anhui Provincial Natural Science Foundation (2508085QA025)
- Anhui Postdoctoral Scientific Research Program Foundation (2024C997)
- China Postdoctoral Science Foundation (2023M733545)
- National Natural Science Foundation of China (12505267)
- National Research Foundation Singapore
This publication has 46 references indexed in Scilit:
- Towards practical reinforcement learning for tokamak magnetic controlFusion Engineering and Design, 2024
- Implementation and design of the control system for poloidal field power supplyFusion Engineering and Design, 2020
- An overview of the numerical methods for tokamak plasma equilibrium computation implemented in the NICE codeFusion Engineering and Design, 2020
- A Comprehensive Survey of Loss Functions in Machine LearningAnnals of Data Science, 2020
- System Identification for EAST plasma shape and position controlFusion Engineering and Design, 2018
- Implementation of GPU parallel equilibrium reconstruction for plasma control in EASTFusion Engineering and Design, 2016
- CREATE-NL+: A robust control-oriented free boundary dynamic plasma equilibrium solverFusion Engineering and Design, 2015
- EAST plasma control systemFusion Engineering and Design, 2008
- Studies of Plasma Equilibrium and Transport in a Tokamak Fusion Device with the Inverse-Variable TechniqueJournal of Computational Physics, 1993
- Dynamic modeling of transport and positional control of tokamaksJournal of Computational Physics, 1986