Before every tokamak startup, a device where superheated gas is held by magnetic fields, operators spend hours calculating how the plasma will behave. Mistakes are too costly: one wrong parameter can ruin the experiment. Now a neural network comes to the rescue, learning from past runs and delivering a prediction faster than you can blink.
The model works like an experienced chef: knowing the recipe (current in the coils, heating power, gas density), it predicts what the dish will turn out like — how hot and stable the plasma will be. Instead of hours of supercomputer calculations, the answer comes in 0.1 seconds. For training, data from 550 discharges of the WEST tokamak in France were used.
Such speed opens new possibilities: you can quickly run through launch scenarios, discard dangerous options, and even adjust parameters in real time during the discharge. Fusion energy is like a complex dance where every step must be precise. The neural network becomes a partner that hints at the next move a moment before the music.
as written by the authors arXiv:2602.19110
Accurately predicting plasma behavior based on discharge configurations is essential for the safe and efficient operation of tokamak experiments. While physics-based integrated modeling codes provide valuable insights, their high computational cost limits their applicability for fast scenario design and control optimization. In this study, we propose a transformer-based machine learning model to predict key global plasma parameters on the WEST tokamak, including the normalized beta ($$\beta_{n}$$), toroidal beta ($$\beta_{t}$$), poloidal beta ($$\beta_{p}$$), plasma stored energy ($$W_{\mathrm{mhd}}$$), safety factor at the magnetic axis ($$q_{0}$$), and safety factor at the 95% flux surface ($$q_{95}$$). The model uses only signals that can be defined before the discharge, such as magnetic coil currents, auxiliary heating power, plasma current reference, and line-averaged plasma density. Trained on 550 discharges from the WEST campaigns, the model demonstrates an average mean square error (MSE) loss of 0.026, an average coefficient of determination $$R^{2}$$ of 0.94, and achieves inference times on the order of 0.1 seconds. These results highlight the potential of data-driven surrogate models for assisting in discharge planning, scenario evaluation, and real-time control of tokamak plasmas.
The WEST tokamak is named after the English word "west" because it is the Western European branch of the international ITER project.