Reconstruction of magnetic configurations in W7-X using artificial neural networks

Böckenhoff, Daniel;Blatzheim, Marko;Hölbe, Hauke;Niemann, Holger;Pisano, Fabio;Labahn, Roger;Pedersen, Thomas Sunn

2018-01-01

---

Abstract

It is demonstrated that artificial neural networks can be used to accurately and efficiently predict details of the magnetic topology at the plasma edge of the Wendelstein 7-X stellarator, based on simulated as well as measured heat load patterns onto plasma-facing components observed with infrared cameras. The connection between heat load patterns and the magnetic topology is a challenging regression problem, but one that suits artificial neural networks well. The use of a neural network makes it feasible to analyze and control the plasma exhaust in real-time, an important goal for Wendelstein 7-X, and for magnetic confinement fusion research in general.