Magnetic Control of WEST Plasmas Through Deep Reinforcement Learning

Tokamaks require magnetic control across a wide range of plasma scenarios. The coupled behavior of plasma dynamics makes deep learning a suitable candidate for efficient control to fulfill these high-dimensional and nonlinear situations. For example, on TCV, deep reinforcement learning (RL) has already been used for tracking of the plasma’s magnetic equilibrium. In this work, we apply such methods to the WEST tokamak, to address control of the plasma’s shape, position, and current, in several relevant configurations. To this end, we developed a distributed framework to train an actor–critic agent on a C++ free-boundary equilibrium code called NICE, in which resistive diffusion allows a more representative evolution of current profile throughout the simulation. The interface between components was done through UDS protocols for fast, asynchronous, and reliable communication. The implemented tool handles feedback control of quantities of interest, with results showing flexibility of the method regarding the use of different training environments.