Speaker
D. D. Carvalho
Description
See the full Abstract at http://ocs.ciemat.es/EPS2018ABS/pdf/P4.1005.pdf
Regularization extraction for real-time plasma tomography at JET
D. R. Ferreira1 , D. D. Carvalho1 , P. J. Carvalho1 , H. Fernandes1 , and JET Contributors∗
EUROfusion Consortium, JET, Culham Science Centre, Abingdon, OX14 3DB, UK
1 Instituto de Plasmas e Fusão Nuclear, Instituto Superior Técnico, Lisbon, Portugal
Plasma tomography [1] consists in reconstructing the 2D radiation profile on a poloidal
cross section of the fusion device. Such reconstruction is based on measurements of the line-
integrated radiation along multiple lines of sight. However, since these lines are sparse, the prob-
lem is under-determined, and a solution must be found by employing regularization methods [2].
Typically, such regularization is implemented by enforcing smoothness along the magnetic flux
surfaces [3] or by using prior knowledge from other diagnostics [4]. In addition, an iterative
procedure is needed to find the optimal regularization parameters [5]. As a result, tomographic
reconstructions are computationally expensive and have extensive knowledge embedded.
In this work, we describe an approach to extract the reg-
KB5V line 4
ularization from existing reconstructions. Based on a set of
2.0
about 500 reconstructions that have been carefully curated at
1.5
JET, we use a machine learning framework to extract a reg-
ularization matrix that can be used to generate new recon- 1.0
structions directly from measurement data. Figure 1 shows
0.5
one of the regularization patterns that can be found in such
Z (m)
0.0
matrix, where a measurement taken along a line of sight con-
tributes to pixels along a curvature that resembles a magnetic 0.5
flux surface. Once the regularization matrix has been found,
1.0
new reconstructions can be computed in a single matrix mul-
1.5
tiplication step, thus enabling the use of plasma tomography
as a real-time diagnostic. 2.0 2.5 3.0 3.5 4.0
R (m)
References
[1] L. C. Ingesson et al., Fusion Sci. Technol. 53, 2 (2008) Figure 1: Regularization pattern
[2] J. Mlynar et al., Fusion Sci. Technol. 58, 3 (2010) for line 4 of the vertical camera
[3] M. Odstrcil et al., Nucl. Instr. Meth. Phys. Res. A 686 (2012)
[4] J. Bielecki et al., Rev. Sci. Instrum. 86, 9 (2015)
[5] V. Loffelmann et al., Fusion Sci. Technol. 69, 2 (2016)
∗
See the author list of X. Litaudon et al., Nuclear Fusion 57, 10 (2017)
This work has been carried out within the framework of the EUROfusion Consortium and has received funding
from the Euratom research and training programme 2014-2018 under grant agreement No 633053. The views and
opinions expressed herein do not necessarily reflect those of the European Commission. IPFN activities received
financial support from Fundação para a Ciência e a Tecnologia (FCT) through project UID/FIS/50010/2013.
