Hardware-in-the-loop validation of an MPC architecture for real-time equilibrium control of tokamak plasmas at RFX-mod2

Abstract

This work deals with the real-time implementation of an equilibrium control system based on Model Predictive Control (MPC) for tokamak plasma operation at RFX-mod2. Such an architecture has been shown to address the challenging asymmetric constraints imposed by the one-quadrant power converters supplying the poloidal field coils at RFX-mod2. However, being based on the solution of constrained optimization problems, the real-time performance of such a control approach must be carefully assessed. To this end, a sensitivity study on the MPC parameters that aims at simplifying the optimization problem to be solved in real-time, while guaranteeing the control performance, is first presented. Hardware-in-the-Loop (HIL) simulations are then performed to assess the real-time performance of the controller. Different off-the-shelf solvers are considered when automatically generating the real-time code starting from the Simulink model. The HIL simulations reveal that the choice of the solver has an impact on the execution time, and may make the real-time implementation infeasible for the chosen optimization problem.