核融合の実現に向けて、プラズマの計測および制御が必要不可欠です。本研究室では静電プローブ計測，レーザーを使ったトムソン散乱計測，分光計測などの手法を用いて特に核融合のプラズマの中でも比較的低温（1000度～10万度程度）のプラズマの計測手法の開発及び詳細な計測を，ダイバータ模擬装置を用いて実施しています。
特に近年，AI技術の進歩によりプラズマ計測に関しても機械学習を取り入れた計測が広く用いられるようになってきました。本研究室では，ニューラルネットワークを使った深層学習によりプラズマからの発光とプラズマ中の温度，密度を学習させて精度よく計測法として利用できる手法を開発してきました。
(AIを使った温度と密度と計測結果の比較。誤差10%程度で温度密度評価が可能になった。Comparison of temperature, density and measurement results using AI technology. Temperature density evaluation is now possible with an error of around 10% [3].）
これらの内容をまとめたレビュー論文を近年執筆しています。
Helium line emission spectroscopy to measure plasma parameters using modeling and machine learning in low-temperature plasmas, S Kajita, D Nishijima，Journal of Physics D: Applied Physics 57 (42), 423003
Plasma measurement and control are indispensable to realize nuclear fusion. In our laboratory, we are developing measurement methods for relatively low temperature plasmas (0.1-10 eV) in the boundary region of fusion reactors using electrostatic probing, Thomson scattering using lasers, and spectroscopy, and are performing detailed measurements using a divertor simulator.
In recent years, in particular, advances in AI technology have led to the widespread use of machine learning in plasma measurement. In this laboratory, we have developed a method that uses deep learning with neural networks to learn the light emitted from plasma and the temperature and density of plasma, and can be used as a highly accurate measurement method.
These contents are summarized in the following review paper.
Helium line emission spectroscopy to measure plasma parameters using modeling and machine learning in low-temperature plasmas, S Kajita, D Nishijima，Journal of Physics D: Applied Physics 57 (42), 423003
関連論文 (Related publication)
[1] S Kajita, D Nishijima, K Fujii, G Akkermans, H van der Meiden
Application of multiple regression for sensitivity analysis of helium line emissions to the electron density and temperature in Magnum-PSI
Plasma Physics and Controlled Fusion 63 (2021), 055018
[2] S Kajita, H Ohshima, H Tanaka, M Seki, H Takano, N Ohno
Spatial and temporal measurement of recombining detached plasmas by laser Thomson scattering
Plasma Sources Science and Technology 28 (2019), 105015
[3] S Kajita, S. Iwai, H. Tanaka, et al.
Use of machine learning for a helium line intensity ratio method in Magnum-PSI
Nuclear Materials and Energy 33 (2022), 101281