Artificial Intelligence for Fusion Reactor Predictions

Fusion research will soon enter the era of reactor-scale facilities as ITER is scheduled for its first plasma in a few years. This project, DEEPfusion, will harness the advancement of artificial intelligence (AI) algorithms to develop solutions for bridging the gap of predictive capability for one of the primary challenges faced by reactor-scale fusion devices: integration of high fusion performance with sustainable heat loads at reactor components. Due to the complexity of fusion plasma physics, predicting the operational space of integrated high performance scenarios with appropriate power exhaust measures in reactor-scale fusion devices is a scientific grand challenge. Traditional scientific approaches are unlikely to fully resolve this challenge within the time frame needed for operation of fusion power plants. DEEPfusion will use state-of-the-art AI algorithms to develop data-driven approaches to solve this scientific grand challenge.