CPS-FR: Magnetic control of TCV tokamak plasmas through Deep Reinforcement Learning

Federico Felici

Swiss Plasma Center, EPFL

Friday, August 26, 2022



PSFC Seminars

A key challenge in tokamak operations is to shape and maintain a high-temperature plasma within the vessel. This requires regulating the plasma position and shape via magnetic fields generated by a set of control coils. This work presents a new architecture for designing a tokamak magnetic controller based on deep reinforcement learning. The controller is entirely trained on a physics-based simulator and then deployed on the TCV tokamak hardware, where it was successful in controlling a diverse set of plasma configurations, including a new configuration featuring two plasmas in the vessel simultaneously. The control architecture replaces separate controllers used in traditional architectures with a single control policy. This lecture will provide details about the training and deployment of the reinforcement learning algorithm, as well as providing a comparison with more traditional control engineering solutions to the magnetic control problem.

Join Zoom Meeting

Password: 843110

One tap mobile
+16465588656,,97529245485# US (New York)
+16699006833,,97529245485# US (San Jose)

Meeting ID: 975 2924 5485

US : +1 646 558 8656 or +1 669 900 6833

International Numbers: https://mit.zoom.us/u/amNCLxJ0l

Join by SIP

Join by Skype for Business