Experimental validation of a fast control-oriented, physics-informed surrogate model for plasma equilibrium reconstruction in the TCV tokamak

Researchers have developed a fast, physics-informed neural network surrogate for plasma equilibrium reconstruction in the TCV tokamak.

The model, trained on thousands of TCV discharges, achieves inference times under 100 microseconds, enabling 10 kHz plasma shape control.

Validated both offline and in real-time, the surrogate demonstrates comparable accuracy to existing methods and has been deployed on the TCV real-time control system.