Preliminary studies on real-time polarization control for DTT’s ECRH system

Electron Cyclotron Resonance Heating (ECRH) is a well-established technique for plasma heating and control in magnetic confinement fusion devices. In this framework, the Divertor Tokamak Test (DTT) facility will deploy one of the world’s largest ECRH system upon completion, comprising 32 gyrotrons operating at the fundamental electron cyclotron frequency of 170 GHz and injecting microwave power from the low-field side (LFS). Each beamline, through which microwave beams are carried, incorporates a pair of rotating grooved polarizing mirrors, located just before the launchers, to maximize power absorption and protect in-vessel components from stray radiation. However, the mapping between polarizing mirrors’ orientation and optimal power coupling exhibits strong nonlinearities and is further affected by unmodeled uncertainties in launch trajectories and edge plasma conditions. Consequently, data-driven approaches become essential to achieve optimal coupling and to mitigate hazardous stray radiation. This work regards the design and development of a simulator for real-time polarization control. An ad-hoc plasma model has been built and validated to mimic the plasma-wave coupling behaviour, providing a stray radiation proxy as output. Four control algorithms have been developed, with a two-level structure that decouples estimation from actuation. The first block, an estimator, computes the gradient of the stray radiation with respect to each polariser rotation angle by exploiting perturbations in the polarisers’ angular positions and using stray radiation measurements. The second block, a controller, is designed to compute the new references for the polarizers’ angles given the gradient of the stray. Among these four algorithms, the Kalman Filter approach for gradient estimation has been selected as the most promising and reliable . Actual tests on operating machines must be performed to determine the real control performance.