Artificial Neural Network (ANN) modeling of the pulsed heat load during ITER CS magnet operation

• First application of Artificial Neural Networks to ITER magnets.
• Pulsed heat load of CS coil to LHe bath studied.
• Predictive capabilities tested successfully in a nominal plasma operating scenario.
• Predictions are accurate and fast.

Artificial Neural Networks (ANNs) are applied to the development of a simplified transient model of the ITER Central Solenoid (CS), aiming at predicting the evolution of the pulsed heat load from the CS to the LHe bath during plasma operation. The ANNs are trained using the thermal–hydraulic evolution in the CS, computed with the 4C code, due to AC losses. The capability of the ANN model to predict the heat load to the LHe bath is successfully demonstrated in the case of different transients, among which a nominal plasma operating scenario. The gain in speed of the simplified model with respect to the 4C code results is by order of magnitudes, with a small loss of accuracy.