Model Predictive Control of the Current Density Distribution and Stored Energy in Tokamak Fusion Experiments using Trajectory Linearizations

Tokamaks are used to confine high temperature plasmas for nuclear fusion research.In this work we apply model predictive control to the transport process in a tokamak plasma that can be described by a set of nonlinear coupled partial differential equations, where the controlled quantities are the current density distribution and stored thermal energy. Applying trajectory linearizations around already commonly predefined feedforward trajectories enables us to use linear MPC techniques that are computationally tractable for implementation on existing tokamaks. Special requirements for the MPC controller are that it should be able to handle real-time-varying references and constraints, whereas the system size, required prediction horizon and available computational time imposes additional challenges. An MPC controller is designed according to the requirements and its performance is analyzed in simulations that approach high performance plasma experiments in the ASDEX Upgrade tokamak. The results show the potential of the controller and encourage its further exploration and use in experiments.