A5031024072- Magnetic confinement fusion research
- Ionosphere and magnetosphere dynamics
- Superconducting Materials and Applications
- Fusion materials and technologies
- Particle accelerators and beam dynamics
- Laser-Plasma Interactions and Diagnostics
- Nuclear reactor physics and engineering
- Nuclear Physics and Applications
Korea Institute of Fusion Energy
2022-2025
Seoul National University
2020-2025
In this work, we address a new feedforward control scheme of the normalized beta (βN) in tokamak plasmas, using deep reinforcement learning (RL) technique. The RL algorithm optimizes an artificial decision-making agent that adjusts discharge scenario to obtain given target βN, from state-action-reward sets explored by trials and errors itself virtual environment. environment for training is constructed with LSTM network imitates plasma responses external actuator controls, which trained...
We report the status of hybrid scenario experiments in Korea Superconducting Tokamak Advanced Research (KSTAR). The is defined as stationary discharges with B N >= 2.4 and H_89 2.0 at q_95 < 6.5 without or very mild sawtooth activities KSTAR. It being developed towards reactor-relevant conditions. High performance
Abstract This work develops an artificially intelligent (AI) tokamak operation design algorithm that provides adequate trajectory to control multiple plasma parameters simultaneously into different targets. An AI is trained with the reinforcement learning technique in data-driven simulator, searching for best action policy get a higher reward. By setting reward function increase as achieved β p , q 95 and l i are close given target values, tries properly determine current boundary shape...
In tokamaks, it is commonly observed that the application of resonant magnetic perturbations (RMPs) leads to a reduction in plasma density. this study, we show decrease density accompanied by kink-like modes edge region KSTAR. The dynamics these toroidal and poloidal directions using multiple diagnostics. It captured phase aligns with applied RMPs. particular, nonuniform surface displacement due measured along direction novel image processing technique on in-vessel TV data. symmetry-breaking...
A newly developed integrated suite of codes coined as tokamak reactor automated for simulation and computation (TRIASSIC) is reported. The comprises existing plasma codes, including 1.5D/2D transport solvers neoclassical/anomalous transport, heating/cooling, cold neutral models. components in TRIASSIC are fully modularized by adopting a generic data structure its internal storage. Primary such the solver beam or electron cyclotron wave actuator were verified to standalone implementation....
Abstract A tokamak, a torus-shaped nuclear fusion device, needs an electric current in the plasma to produce magnetic field poloidal direction for confining plasmas. Plasma is conventionally generated by electromagnetic induction. However, steady-state reactor, minimizing inductive essential extend tokamak operating duration. Several non-inductive drive schemes have been developed operations such as radio-frequency waves and neutral beams. commercial reactors require minimal use of these...
Abstract We report experimental observations on the effect of plasma boundary shaping towards balanced double-null (DN) configuration performance in KSTAR. The transition from a single-null to DN resulted improved performance, manifested through changes pedestal region, decreased density, and core MHD activity variation. Specifically, led wider higher structure, accompanied by grassy edge-localized modes (ELMs) characteristics. density decrease was prerequisite for enhancement during...
Abstract This paper deals with one of the origins and trigger mechanisms responsible for observed performance enhancements in hybrid scenario experiments conducted Korea Superconducting Tokamak Advanced Research (KSTAR). The major contribution to improvement comes from a broader higher pedestal formation. increase fast ion pressure due plasma density decrease also contributes substantially global beta. Although reduced core volume resulting expansion has negative effect on thermal energy,...
Abstract The neural network model, MISHKA-NN is developed to mitigate the computational burden associated with linear ideal magnetohydrodynamic (MHD) stability analysis of pedestal based on peeling–ballooning (P–B) model. By utilizing both 1D plasma profiles (current density, pressure gradient, and safety factor) 0D parameters (plasma geometry, total current, toroidal mode number), model predicts growth rate edge-localized MHD instability in a given equilibrium state. enabling prediction...
Classifying and monitoring the L-, H-mode, plasma-free state are essential for stable operational control of tokamaks. Edge reflectometry measures plasma density profiles, but large volume data complexity in reconstruction pose significant challenges. There is a need efficient methods to analyze complex reflectometer real-time, which can be addressed using advanced computational techniques. Here, we show that machine learning (ML) techniques classify discharge states raw signal from an edge...
The new neural-network model for the pedestal linear MHD (magnetohydrodynamic) stability analysis is developed, to accelerate speed and reduce numerical burden. This predicts growth rates of edge-localized instabilities KSTAR-like (Korea Superconducting Tokamak Advanced Research) plasma with a structure at edge. trained by data set consisting parametric equilibria calculation results. It has successfully predicted most unstable toroidal mode number benchmark cases within reasonable errors....
Abstract We report a discovery of fusion plasma regime suitable for commercial reactor where the ion temperature was sustained above 100 million degree about 20 s first time. Nuclear as promising technology replacing carbon-dependent energy sources has currently many issues to be resolved enable its large-scale use sustainable source. State-of-the-art reactors cannot yet achieve high levels performance, temperature, and absence instabilities required steady-state operation long period time...