Plasma initiation is an important phase in a tokamak discharge and its design and optimization is getting more and more attention in view of the operation of large tokamaks like ITER. The main objective of magnetic control during this phase is to obtain a high electric field to ionize the neutral particles with a low stray magnetic field to avoid the ionized particles escaping towards the chamber walls, in a sufficiently large region inside the vacuum chamber, and then, sustain the plasma current rise whilst maintaining the force balance equilibrium. This paper describes the application of a recent plasma initiation optimisation algorithm, implemented in the CREATE-BD code, to the MAST Upgrade (MAST-U) tokamak. The procedure is based on quadratic programming and iterative learning control methodologies. In fact the breakdown scenario is corrected step by step on the basis of the previous experiments converging to an optimal solution in few steps.

In this paper we propose a modular approach, based on a deep reinforcement learning technique, for the control of a plasma with a limited configuration in the DEMO tokamak. Three different reinforcement learning agents are used to perform the magnetic confinement of the plasma, i.e. to stabilize the vertical plasma instability, to control the radial centroid position, and to ramp-up the plasma current. This modular approach allows us to simplify the training procedure of the control policy, since it requires a lower overall computational load. Performance of the proposed approach are characterized by numerical simulations.

In order to test aircraft actuators moving aerodynamic control surfaces, it becomes more and more important to have flexible facilities which are able to produce a variety of loading conditions with a given accuracy. The so-called Modular Iron Bird (MIB), which is now in the design phase, will be able to reproduce aerodynamic and inertial loads for a large category of aircraft. The MIB control system has to guarantee the application of a desired time history of forces through a hydraulic system, while the aircraft actuator is driven in closed-loop to guarantee a given aerodynamic surface deflection. Therefore, two controllers are fighting each other, and this makes the control problem challenging. In this paper, a nonlinear model-based predictive control strategy is proposed. Numerical results allow to test the effectiveness of this control strategy and its benefit with respect to the classical PID approach.