Bremen 2017 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

P: Fachverband Plasmaphysik

P 14: Plasma Wall Interaction

P 14.6: Poster

Dienstag, 14. März 2017, 16:30–18:30, HS Foyer

Protection of the first wall of Wendelstein 7-X with neural nets. — •Daniel Böckenhoff and Marko Blatzheim — Max-Planck-Institut für Plasmaphysik, Greifswald, Deutschland

Wendelstein 7-X (W7-X) is the first fully optimized stellarator. One of its main objectives is to demonstrate steady state capability of the confinement concept. The plasma-wall contact is going to be realized with so called divertors that intersect the plasma. Various mechanisms, like the development of plasma currents, lead to a change in the magnetic topology as well as plasma parameters over time. Therefore the heat load pattern on the divertors is dynamic. To ensure the safety of the first wall and protect the plasma from impurities, heat load pattern control is essential for long term operation.

Since the physics of the underlying processes is highly complex, we seek plasma control based on artificial neural networks. Inputs are in particular infrared pictures of the heat load pattern. The first important step towards this goal is to train the neural net to reconstruct the magnetic edge topology from a series of given patterns and predict the further evolution. Moreover, a study of the optimal input quantities and their preprocessing is crucial and will be presented herein.