Plasma and Fusion Research

Volume 20, 1301039 (2025)

Letters

Physics-Informed Machine Learning Approach to Modeling Line Emission from Helium-Containing Plasmas
Shin KAJITA
Graduate School of Frontier Sciences, The University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa, Chiba 277-8561, Japan
(Received 12 May 2025 / Accepted 17 June 2025 / Published 13 August 2025)

Abstract

The helium I line intensity ratio (LIR) method is used to measure the electron density (ne) and temperature (Te) of fusion-relevant plasmas. Although the collisional-radiative model (CRM) has been used to predict ne and Te, recent studies have shown that machine learning approaches can provide better measurements if a sufficient dataset for training is available. This study investigates a hybrid neural network approach that combines CRM- and experiment-based models. Although the CRM-based model alone exhibited negative transfer in most cases, the ensemble model modestly improved the prediction accuracy of Te. Notably, in data-limited scenarios, the CRM-based model outperformed the others for Te prediction, highlighting its potential for applications with constrained diagnostic access.

Keywords

optical emission spectroscopy, helium line intensity ratio method, diagnostics, collisional-radiative model, artificial intelligence

DOI: 10.1585/pfr.20.1301039

Full Text

PDF format (1.0Mbytes) PDF Download

References

  • [1] B. Schweer et al., J. Nucl. Mater. 196–198, 174 (1992).
  • [2] M. Agostini et al., Rev. Sci. Instrum. 91, 113503 (2020).
  • [3] E. de la Cal, Plasma Phys. Control. Fusion 43, 813 (2001).
  • [4] N. Ohno et al., Contrib. Plasma Phys. 50, 962 (2010).
  • [5] S. Kimata et al., AIP Advances 12, 045204 (2022).
  • [6] M. Goto, J. Quant. Spectrosc. Radiat. Transf. 76, 331 (2003).
  • [7] S. Sasaki et al., Rev. Sci. Instrum. 67, 3521 (1996).
  • [8] S. Kajita et al., Phys. Plasmas 13, 013301 (2006).
  • [9] S. Kajita et al., Phys. Plasmas 16, 063303 (2009).
  • [10] S. Kajita et al., Phys. Plasmas 16, 029901 (2009).
  • [11] Y. Iida et al., Rev. Sci. Instrum. 81, 10E511 (2010).
  • [12] K. Sawada et al., Plasma Fusion Res. 5, 001 (2010).
  • [13] D. Nishijima and E.M. Hollmann, Plasma Phys. Control. Fusion 49, 791 (2007).
  • [14] S. Kajita and N. Ohno, Rev. Sci. Instrum. 82, 023501 (2011).
  • [15] Y. Iida et al., Phys. Plasmas 17, 123301 (2010).
  • [16] S. Kajita et al., Phys. Plasmas 25, 063303 (2018).
  • [17] S. Kajita et al., AIP Advances 10, 025225 (2020).
  • [18] D. Nishijima et al., Rev. Sci. Instrum. 92, 023505 (2021).
  • [19] S. Kajita et al., Plasma Phys. Control. Fusion 63, 055018 (2021).
  • [20] S. Kajita et al., Nucl. Mater. Energy 33, 101281 (2022).
  • [21] S. Kajita et al., Fusion Eng. Des. 196, 114012 (2023).
  • [22] G.E. Karniadakis et al., Nat. Rev. Phys. 3, 422 (2021).
  • [23] S. Kajita et al., Plasma Phys. Control. Fusion 66, 105005 (2024).
  • [24] H. van Eck et al., Fusion Eng. Des. 89, 2150 (2014).
  • [25] S. Kajita and D. Nishijima, J. Phys. D: Appl. Phys. 57, 423003 (2024).
  • [26] M. Abadi et al., arXiv preprint arXiv:1603.04467 (2016).
  • [27] F. Chollet et al., keras (2015).
  • [28] W. Zhang et al., IEEE/CAA J. Autom. Sin. 10, 305 (2023).