In a tokamak with superconducting magnets, the operation of the cryoplant requires the knowledge of the heat load coming from the cryogenic loops that cool the different magnet systems.
Artificial Neural Networks (ANNs) are applied for the first time to the ITER Toroidal Field (TF) magnets. Two different models are developed: 1) a simpler one, aiming at checking the effects of the different operating scenarios on the cryoplant; 2) a more complex one, aiming at helping in the design of suitable control strategies for the magnet operation, to reduce the variation of the heat load to the cryoplant.
The developed ANNs are suitably trained based on results obtained with the state-of-the-art thermal-hydraulic code 4C, that simulates the TF magnet response when subject to a broad spectrum of heat load variations. The predictive capability of the resulting ANN models is tested in different operating scenarios.