Jayhyun Kim
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Jayhyun Kim has a total of 106 co-authors. Here are the top co-authors by number of shared publications:
- Jun Gyo Bak
Korea Institute of Fusion Energy
5 shared publications - Sang-Hee Hahn
Korea Institute of Fusion Energy , Daejeon 34133
5 shared publications - Jeongwon Lee
Korea Institute of Fusion Energy
4 shared publications - Hyunsun Han
Korea Institute of Fusion Energy
3 shared publications - Michael Lehnen
ITER Organization
3 shared publications - Young-Chul Ghim
Department of Nuclear and Quantum Engineering Korea Advanced Institute of Science and Technology 291 Daehak‐ro Yuseong‐gu Daejeon 34141 Republic of Korea
3 shared publications
Latest Publications
- Open Access
- 4 November 2026
- Published by IOP Publishing in Nuclear Fusion
- Vol. 67 (1) , 016022
AI-based plasma equilibrium parameter inference in KSTAR for DEMO-relevant reactors using only ex-vessel magnetic diagnostics
This study proposes a deep neural network for high-fidelity equilibrium parameter inference, serving as a surrogate of the magnetic EFIT reconstruction and relying exclusively on ex-vessel magnetic signals. The objective is to circumvent the long-term integrity constraints of in-vessel magnetic diagnostics in the nuclear environment of the Demonstration Fusion Power Reactor (DEMO)-relevant devices. Using KSTAR data (2021–2024), a baseline model trained solely on measured ex-vessel magnetic diagnostics (the Measurement-only model) is compared with an Augmented model, which addresses informational sparsity via virtual-sensor signals computed from the GEQDSK. The Augmented model achieves predictive fidelity comparable to off-line EFIT, with a coefficient of determination ( R 2 ) above 0.99 and a range-normalized root-mean-squared error (NRMSE) of 0.332 % averaged over all parameters, against 2.385 % for the Measurement-only model. Because the virtual-sensor inputs derive from the EFIT solutions that define the targets, these figures quantify the surrogate fidelity with respect to the magnetic EFIT and constitute an upper bound on the accuracy attainable with physically installed ex-vessel probes, not that of an independent reconstruction. As the reference floor for the Measurement-only model, the magnetic fitting residuals of the parent EFIT reconstructions against the measured probe signals are quantified, yielding pooled range-normalized residuals of 1.55 % – 3.87 % per campaign with structured, campaign-persistent components that random noise perturbations do not emulate. Robust accuracy is maintained during neutral beam injection (NBI)-induced transients, and Shapley Additive exPlanations (SHAP)-based attribution verifies that the Augmented model captures the underlying magnetohydrodynamic physics, specifically the Shafranov shift. With a median single forward pass inference latency of 3.47 ms (99th percentile: 3.53 ms), the framework is a viable surrogate for conventional numerical solvers. In conclusion, this study demonstrates that an idealized, EFIT-consistent realization of a densified ex-vessel sensor configuration retains sufficient information for high-precision equilibrium parameter inference, provided that magnetic probes are physically installed at the designated locations. This establishes the information-content prerequisite for ex-vessel-only inference in DEMO. The accuracy attainable from physically installed probes, subject to conductor-model, calibration and structured measurement-model errors, is not established and requires validation against measured ex-vessel signals.
- 4 November 2026
- Published by IOP Publishing in Nuclear Fusion
- Vol. 67 (1) , 016022
Open AccessAI-based plasma equilibrium parameter inference in KSTAR for DEMO-relevant reactors using only ex-vessel magnetic diagnostics
This study proposes a deep neural network for high-fidelity equilibrium parameter inference, serving as a surrogate of the magnetic EFIT reconstruction and relying exclusively on ex-vessel magnetic signals. The objective is to circumvent the long-term integrity constraints of in-vessel magnetic diagnostics in the nuclear environment of the Demonstration Fusion Power Reactor (DEMO)-relevant devices. Using KSTAR data (2021–2024), a baseline model trained solely on measured ex-vessel magnetic diagnostics (the Measurement-only model) is compared with an Augmented model, which addresses informational sparsity via virtual-sensor signals computed from the GEQDSK. The Augmented model achieves predictive fidelity comparable to off-line EFIT, with a coefficient of determination ( R 2 ) above 0.99 and a range-normalized root-mean-squared error (NRMSE) of 0.332 % averaged over all parameters, against 2.385 % for the Measurement-only model. Because the virtual-sensor inputs derive from the EFIT solutions that define the targets, these figures quantify the surrogate fidelity with respect to the magnetic EFIT and constitute an upper bound on the accuracy attainable with physically installed ex-vessel probes, not that of an independent reconstruction. As the reference floor for the Measurement-only model, the magnetic fitting residuals of the parent EFIT reconstructions against the measured probe signals are quantified, yielding pooled range-normalized residuals of 1.55 % – 3.87 % per campaign with structured, campaign-persistent components that random noise perturbations do not emulate. Robust accuracy is maintained during neutral beam injection (NBI)-induced transients, and Shapley Additive exPlanations (SHAP)-based attribution verifies that the Augmented model captures the underlying magnetohydrodynamic physics, specifically the Shafranov shift. With a median single forward pass inference latency of 3.47 ms (99th percentile: 3.53 ms), the framework is a viable surrogate for conventional numerical solvers. In conclusion, this study demonstrates that an idealized, EFIT-consistent realization of a densified ex-vessel sensor configuration retains sufficient information for high-precision equilibrium parameter inference, provided that magnetic probes are physically installed at the designated locations. This establishes the information-content prerequisite for ex-vessel-only inference in DEMO. The accuracy attainable from physically installed probes, subject to conductor-model, calibration and structured measurement-model errors, is not established and requires validation against measured ex-vessel signals.
- Open Access
- 7 October 2026
- Published by IOP Publishing in Nuclear Fusion
- Vol. 66 (12) , 126043
Revised error field assessment in KSTAR with profile diagnostics of resonant field penetration
The assessment of the n = 1 intrinsic error field in KSTAR has been revised in three respects. First, a local penetration timing criterion is defined from electron cyclotron emission (ECE) and charge exchange spectroscopy (CES) signatures near the reconstructed q = 2 surface. This timing determines the in-vessel control coil current used as the penetration threshold in the compass scan analysis. Second, the compass scans with the MID and OFFMID coils in 1.8 T L-mode plasmas indicate an intrinsic EF level up to an order of magnitude larger than the 2015 KSTAR reference value, implying that the intrinsic EF can change over extended operation and following hardware modifications. Third, the inferred EF direction is assessed by comparing the MID and OFFMID results in terms of the General Perturbed Equilibrium Code overlap field, where the two configurations give nearly the same direction despite their different poloidal spectra. A rotating resonant magnetic perturbation (RMP) experiment provides an additional dynamic check of this direction, with penetration observed near the phase predicted from the static MID scan. The reversed- B t comparison further constrains the possible source symmetry and is more compatible with a PF related contribution than with a TF dominated contribution, although the specific hardware source is not identified. These results provide an updated EF reference for KSTAR operation and motivate periodic reassessment of the intrinsic EF over the operational lifetime of the device.
- 7 October 2026
- Published by IOP Publishing in Nuclear Fusion
- Vol. 66 (12) , 126043
Open AccessRevised error field assessment in KSTAR with profile diagnostics of resonant field penetration
The assessment of the n = 1 intrinsic error field in KSTAR has been revised in three respects. First, a local penetration timing criterion is defined from electron cyclotron emission (ECE) and charge exchange spectroscopy (CES) signatures near the reconstructed q = 2 surface. This timing determines the in-vessel control coil current used as the penetration threshold in the compass scan analysis. Second, the compass scans with the MID and OFFMID coils in 1.8 T L-mode plasmas indicate an intrinsic EF level up to an order of magnitude larger than the 2015 KSTAR reference value, implying that the intrinsic EF can change over extended operation and following hardware modifications. Third, the inferred EF direction is assessed by comparing the MID and OFFMID results in terms of the General Perturbed Equilibrium Code overlap field, where the two configurations give nearly the same direction despite their different poloidal spectra. A rotating resonant magnetic perturbation (RMP) experiment provides an additional dynamic check of this direction, with penetration observed near the phase predicted from the static MID scan. The reversed- B t comparison further constrains the possible source symmetry and is more compatible with a PF related contribution than with a TF dominated contribution, although the specific hardware source is not identified. These results provide an updated EF reference for KSTAR operation and motivate periodic reassessment of the intrinsic EF over the operational lifetime of the device.
- 1 September 2026
- Published by AIP Publishing in Review of Scientific Instruments
- Vol. 97 (9)
Development of Korea Superconducting Tokamak Advanced Researches (KSTAR) Lyman alpha diagnostics and its synthetic data algorithm
Characterizing the neutral density in the edge region of the magnetically confined plasma is critical for understanding power exhaust, fuel recycling, and the L–H transition. Lyman-α emission can offer two advantages over Balmer-α: higher emission intensity where the edge electron density is below 1019/m3, and simpler data analysis because the emissivity is directly tied to the ground-state atomic density, with a smaller molecular contribution at low temperatures near a divertor and the magnetic X-point. We report the Korea Superconducting Tokamak Advanced Research (KSTAR) Lyman Alpha Measuring Apparatus (KLAMA), a radial 20-channel system viewing the region near the magnetic X-point and lower divertor, with a radial channel separation of up to 1.3 cm. KLAMA uses three notch reflective filters designed to provide 13.02% transmission while suppressing the neighboring C-III line by a factor of ∼14. We have developed a synthetic data algorithm utilizing the Virtual KSTAR platform to optimize the optical design, quantify noise propagation, and build the transfer matrix. The synthetic data algorithm is also used for a Gaussian-process-regularized 2D reconstruction of the neutral density and is expected to be used to generate a forward model for a Bayesian inference analysis in future work. KLAMA has been manufactured and installed on KSTAR, and preliminary data acquired during a disruption are presented.
- 1 September 2026
- Published by AIP Publishing in Review of Scientific Instruments
- Vol. 97 (9)
Development of Korea Superconducting Tokamak Advanced Researches (KSTAR) Lyman alpha diagnostics and its synthetic data algorithm
Characterizing the neutral density in the edge region of the magnetically confined plasma is critical for understanding power exhaust, fuel recycling, and the L–H transition. Lyman-α emission can offer two advantages over Balmer-α: higher emission intensity where the edge electron density is below 1019/m3, and simpler data analysis because the emissivity is directly tied to the ground-state atomic density, with a smaller molecular contribution at low temperatures near a divertor and the magnetic X-point. We report the Korea Superconducting Tokamak Advanced Research (KSTAR) Lyman Alpha Measuring Apparatus (KLAMA), a radial 20-channel system viewing the region near the magnetic X-point and lower divertor, with a radial channel separation of up to 1.3 cm. KLAMA uses three notch reflective filters designed to provide 13.02% transmission while suppressing the neighboring C-III line by a factor of ∼14. We have developed a synthetic data algorithm utilizing the Virtual KSTAR platform to optimize the optical design, quantify noise propagation, and build the transfer matrix. The synthetic data algorithm is also used for a Gaussian-process-regularized 2D reconstruction of the neutral density and is expected to be used to generate a forward model for a Bayesian inference analysis in future work. KLAMA has been manufactured and installed on KSTAR, and preliminary data acquired during a disruption are presented.
- Open Access
- 17 August 2026
- Published by Elsevier in Fusion Engineering and Design
- Vol. 232
FEM-based plasma equilibrium reconstruction for transient and off-normal phases in KSTAR
- 17 August 2026
- Published by Elsevier in Fusion Engineering and Design
- Vol. 232
Open AccessFEM-based plasma equilibrium reconstruction for transient and off-normal phases in KSTAR
- Open Access
- 1 March 2026
- Published by AIP Publishing in Physics of Plasmas
- Vol. 33 (3)
Global energy balance simulations of shattered pellet injection
The dynamics of shattered pellet injection (SPI) shutdowns are simulated using a time-dependent global energy balance model, based on a modification of the KPRAD framework. The new SPI particle source in the model calculates the ablation of individual pellet fragments that enter the plasma as a temporally resolved plume, thus capturing the effects of earlier fragments on the ablation of those that follow, which has a significant impact on the overall assimilation. Despite the reduced physics and the global averaging of all quantities, results from a large number of DIII-D, KSTAR, and JET experiments are well reproduced, including the plasma cooling timescales, particle assimilation, and current quench (CQ) rates. Cooling timescales and CQ rates are in good agreement for pellets containing as little as ∼1% neon, while particle assimilations are most accurate for neon fractions above ∼15% by number of atoms. Below this, the assimilation tends to be overestimated due to the lack of radial transport in the particle balance, which becomes important in the low-Z limit. Predictive simulations of mixed-composition dual-SPI shutdowns in ITER are compared against those with the 3D non-linear magnetohydrodynamic code JOREK and are found to reproduce overall trends observed in the higher-fidelity modeling across a range of injection scenarios. The general success of the model points to the critical role of energy balance in determining SPI particle assimilation and the subsequent disruption dynamics and highlights the value of these simulations for experimental interpretation and for optimizing the deployment of computationally expensive, higher-fidelity models.
- 1 March 2026
- Published by AIP Publishing in Physics of Plasmas
- Vol. 33 (3)
Open AccessGlobal energy balance simulations of shattered pellet injection
The dynamics of shattered pellet injection (SPI) shutdowns are simulated using a time-dependent global energy balance model, based on a modification of the KPRAD framework. The new SPI particle source in the model calculates the ablation of individual pellet fragments that enter the plasma as a temporally resolved plume, thus capturing the effects of earlier fragments on the ablation of those that follow, which has a significant impact on the overall assimilation. Despite the reduced physics and the global averaging of all quantities, results from a large number of DIII-D, KSTAR, and JET experiments are well reproduced, including the plasma cooling timescales, particle assimilation, and current quench (CQ) rates. Cooling timescales and CQ rates are in good agreement for pellets containing as little as ∼1% neon, while particle assimilations are most accurate for neon fractions above ∼15% by number of atoms. Below this, the assimilation tends to be overestimated due to the lack of radial transport in the particle balance, which becomes important in the low-Z limit. Predictive simulations of mixed-composition dual-SPI shutdowns in ITER are compared against those with the 3D non-linear magnetohydrodynamic code JOREK and are found to reproduce overall trends observed in the higher-fidelity modeling across a range of injection scenarios. The general success of the model points to the critical role of energy balance in determining SPI particle assimilation and the subsequent disruption dynamics and highlights the value of these simulations for experimental interpretation and for optimizing the deployment of computationally expensive, higher-fidelity models.
- Open Access
- 5 March 2025
- Published by IOP Publishing in Nuclear Fusion
- Vol. 65 (4) , 046008
Sensor fusion and magnetic drift estimation in magnetic measurements using Kalman filter for long-duration plasma operations
This work introduces a novel technique that incorporates the bias contaminating magnetic coil measurements within the model, building upon previous sensor fusion techniques of magnetic coil and Hall sensor (Quercia et al 2022 Nucl. Fusion 62 106032; Arpaia et al 2021 Sensors 22 182). In the pursuit of sustainable thermo-nuclear magnetic fusion, precise and drift-free magnetic field measurements are essential for effective plasma control. Typically, magnetic field measurements for plasma control and diagnosis in magnetic fusion are achieved through inductive coil sensors and integrators, striving for low noise and fast sampling rates. However, these methods are susceptible to drift due to the presence of bias or offset originated from the magnetic coil d B / d t measurements and the integrator. To address the drift issue, we employ a Kalman filter approach that considers the time-varying bias as a Wiener or Brownian process, in conjunction with a Hall sensor. This method is applied to synthetic magnetic field data representing Hall sensor measurements with high noise and synthetic d B / d t data representing magnetic coil sensor measurements with time-varying bias. While these conditions are more extreme than typical in current fusion diagnostics, they are chosen to rigorously test the robustness of our method. The results demonstrate successful reconstruction of a low-noise, low-bias magnetic field and the time-varying bias, highlighting the method’s reliability in challenging scenarios. The proposed method offers promising applications in achieving long-term, drift-free control of plasma in magnetic fusion experiments, taking us one step closer to the goal of sustainable fusion energy production.
- 5 March 2025
- Published by IOP Publishing in Nuclear Fusion
- Vol. 65 (4) , 046008
Open AccessSensor fusion and magnetic drift estimation in magnetic measurements using Kalman filter for long-duration plasma operations
This work introduces a novel technique that incorporates the bias contaminating magnetic coil measurements within the model, building upon previous sensor fusion techniques of magnetic coil and Hall sensor (Quercia et al 2022 Nucl. Fusion 62 106032; Arpaia et al 2021 Sensors 22 182). In the pursuit of sustainable thermo-nuclear magnetic fusion, precise and drift-free magnetic field measurements are essential for effective plasma control. Typically, magnetic field measurements for plasma control and diagnosis in magnetic fusion are achieved through inductive coil sensors and integrators, striving for low noise and fast sampling rates. However, these methods are susceptible to drift due to the presence of bias or offset originated from the magnetic coil d B / d t measurements and the integrator. To address the drift issue, we employ a Kalman filter approach that considers the time-varying bias as a Wiener or Brownian process, in conjunction with a Hall sensor. This method is applied to synthetic magnetic field data representing Hall sensor measurements with high noise and synthetic d B / d t data representing magnetic coil sensor measurements with time-varying bias. While these conditions are more extreme than typical in current fusion diagnostics, they are chosen to rigorously test the robustness of our method. The results demonstrate successful reconstruction of a low-noise, low-bias magnetic field and the time-varying bias, highlighting the method’s reliability in challenging scenarios. The proposed method offers promising applications in achieving long-term, drift-free control of plasma in magnetic fusion experiments, taking us one step closer to the goal of sustainable fusion energy production.
- 22 October 2024
- Published by Elsevier in Fusion Engineering and Design
- Vol. 208
Startup experiment with the newly installed lower tungsten divertor of KSTAR
- 22 October 2024
- Published by Elsevier in Fusion Engineering and Design
- Vol. 208
Startup experiment with the newly installed lower tungsten divertor of KSTAR
- Open Access
- 24 September 2024
- Published by IOP Publishing in Nuclear Fusion
- Vol. 64 (12) , 129601
Erratum: Nonlinear MHD modeling of neon doped shattered pellet injection with JOREK and its comparison to experiments in KSTAR (2024 Nucl. Fusion 64 106042)
- 24 September 2024
- Published by IOP Publishing in Nuclear Fusion
- Vol. 64 (12) , 129601
Open AccessErratum: Nonlinear MHD modeling of neon doped shattered pellet injection with JOREK and its comparison to experiments in KSTAR (2024 Nucl. Fusion 64 106042)
- Open Access
- 4 September 2024
- Published by IOP Publishing in Nuclear Fusion
- Vol. 64 (10) , 106042
Nonlinear MHD modeling of neon doped shattered pellet injection with JOREK and its comparison to experiments in KSTAR
3D nonlinear MHD simulations of neon-doped single shattered pellet injection (SPI) conducted with the JOREK code reveal rich physics during SPI-induced disruptions in KSTAR. In the early phase, pressure-driven modes dominate, and the perturbation of the plasma current is largely consistent with the perturbation of the Pfirsch–Schlüter current. As shards reach the q = 1 surface, resistive current perturbations by helical electron cooling start to dominate, and the electron temperature in the core begins to collapse with convective mixing of the density driven by the internal 1 / 1 kink mode. The confinement of the plasma is later completely destroyed as a cold bubble convects into the core via quasi-interchange mode. Comparisons with available experimental data demonstrate qualitative agreements between JOREK results and experiments, and possible reasons for deviations are discussed.
- 4 September 2024
- Published by IOP Publishing in Nuclear Fusion
- Vol. 64 (10) , 106042
Open AccessNonlinear MHD modeling of neon doped shattered pellet injection with JOREK and its comparison to experiments in KSTAR
3D nonlinear MHD simulations of neon-doped single shattered pellet injection (SPI) conducted with the JOREK code reveal rich physics during SPI-induced disruptions in KSTAR. In the early phase, pressure-driven modes dominate, and the perturbation of the plasma current is largely consistent with the perturbation of the Pfirsch–Schlüter current. As shards reach the q = 1 surface, resistive current perturbations by helical electron cooling start to dominate, and the electron temperature in the core begins to collapse with convective mixing of the density driven by the internal 1 / 1 kink mode. The confinement of the plasma is later completely destroyed as a cold bubble convects into the core via quasi-interchange mode. Comparisons with available experimental data demonstrate qualitative agreements between JOREK results and experiments, and possible reasons for deviations are discussed.
- Open Access
- 7 September 2023
- Published by IOP Publishing in Nuclear Fusion
- Vol. 63 (10) , 106017
Gaussian process-based quasi-coherent noise suppression in magnetic confinement devices with superconductors
Gaussian process (GP)-based technique suppressing quasi-coherent noises, i.e. structured noises, is developed which is more effective than conventional denoising techniques such as using frequency-domain filters. Superconducting devices like KSTAR, EAST, JT-60SA and ITER require separate sets of normal conducting magnetic coils inside the tokamak vacuum vessels to achieve a prompt control of fusion-grade plasmas in response to various fast and abrupt plasma activities such as vertical displacement events. Hence, these in-vessel control coils are typically operated with high-frequency switching power supplies which generate quasi-coherent noises. Semi-conductor based bolometers in KSTAR, for instance, are vulnerable to the quasi-coherent noise that makes a tomographic reconstruction for the 2D poloidal radiation map with the noise-contaminated signals flawed. By modeling the quasi-coherent properties of the noise as multivariate Gaussian distribution and generating the kernel function for the GP solely based on the measurements, the proposed method is able to suppress the noise whose performance is superior to the conventional filtering schemes. The method not only suggests an estimate of the denoised signal but also informs the consistent (with the measurements) uncertainty of the estimate at a level smaller than the standard deviation of the quasi-coherent noise. Performance of the method is confirmed with synthetic data containing the quasi-coherent noises, and it is applied to the measured data obtained by the KSTAR bolometers.
- 7 September 2023
- Published by IOP Publishing in Nuclear Fusion
- Vol. 63 (10) , 106017
Open AccessGaussian process-based quasi-coherent noise suppression in magnetic confinement devices with superconductors
Gaussian process (GP)-based technique suppressing quasi-coherent noises, i.e. structured noises, is developed which is more effective than conventional denoising techniques such as using frequency-domain filters. Superconducting devices like KSTAR, EAST, JT-60SA and ITER require separate sets of normal conducting magnetic coils inside the tokamak vacuum vessels to achieve a prompt control of fusion-grade plasmas in response to various fast and abrupt plasma activities such as vertical displacement events. Hence, these in-vessel control coils are typically operated with high-frequency switching power supplies which generate quasi-coherent noises. Semi-conductor based bolometers in KSTAR, for instance, are vulnerable to the quasi-coherent noise that makes a tomographic reconstruction for the 2D poloidal radiation map with the noise-contaminated signals flawed. By modeling the quasi-coherent properties of the noise as multivariate Gaussian distribution and generating the kernel function for the GP solely based on the measurements, the proposed method is able to suppress the noise whose performance is superior to the conventional filtering schemes. The method not only suggests an estimate of the denoised signal but also informs the consistent (with the measurements) uncertainty of the estimate at a level smaller than the standard deviation of the quasi-coherent noise. Performance of the method is confirmed with synthetic data containing the quasi-coherent noises, and it is applied to the measured data obtained by the KSTAR bolometers.