Physics-Informed Deep Learning for Plasma Turbulence Predictions in Fusion Reactors

In this project, the aim is to develop next-generation artificial intelligence (AI) methods to enhance the operation of fusion reactors. Fusion reactors are carbon-free energy sources that produce energy with the same physical phenomenon as stars do. From AI, we use so-called physics-informed machine learning methods which aim at improving the efficiency of the training of the AI methodology by reducing the need for training data by incorporating physical models into the machine learning model. The operation of the reactor is improved by more accurate modeling and prediction of plasma edge turbulence.