Optimal MSE polarisation angle and q-profile estimation using Kalman filters and the plasma simulator RAPTOR

EUROfusion MST1 Team, ASDEX-Upgrade team, M.C.C. Messmer (Corresponding author)

Research output: Contribution to journalArticleAcademicpeer-review

2 Citations (Scopus)
71 Downloads (Pure)

Abstract

Accurate q-profile reconstruction is of importance for the development of advanced scenarios, but continues to be a challenge in tokamak research. To constrain the q-profile in the plasma centre the motional Stark effect diagnostic (MSE) is often used, however achieving routine measurements with the required accuracy proves to be difficult in many devices. We present a novel approach to obtain accurate estimates of the q-profile using an observer based approach. The observer combines MSE measurements with a model-based prediction of the system. For this the plasma transport simulator RAPTOR is coupled with a fixed-boundary equilibrium solver to create a model-based prediction of the MSE measurements. An extended Kalman filter is used to merge profile evolution predictions from the RAPTOR code with measurements. Using synthetic data we demonstrate accurate q-profile estimations in situations where the model is purposely disturbed and only erroneous MSE measurements are available. For shots at ASDEX Upgrade we show that by constraining RAPTOR with MSE measurements, the evolution of the model's q-profile is in close proximity to reference profiles of reconstructed equilibria from an integrated diagnostic suite.

Original languageEnglish
Article number035011
Number of pages11
JournalPlasma Physics and Controlled Fusion
Volume61
Issue number3
DOIs
Publication statusPublished - 1 Feb 2019

Keywords

  • extended Kalman filter
  • motional Stark effect
  • MSE
  • observer
  • q-profile reconstruction
  • RAPTOR
  • tokamak control

Fingerprint Dive into the research topics of 'Optimal MSE polarisation angle and q-profile estimation using Kalman filters and the plasma simulator RAPTOR'. Together they form a unique fingerprint.

Cite this