Abstract
Neural networks have been implemented to reconstruct electron temperature profiles from multi-energy soft-x-ray (ME-SXR) arrays and other plasma diagnostics with fast time resolution. On NSTX, electron temperature profiles are measured with a Thomson scattering diagnostic at 60 Hz, a speed limited by the repetition rate of the lasers. By training a neural network to match fast (>10 kHz) x-ray data with Te profiles from Thomson scattering, the ME-SXR diagnostic can be used to produce Te profiles with high time resolution. In particular, a new ME-SXR system will be used in conjunction with a new laser blow-off impurity injection system to measure cold pulse propagation in NSTX-U plasmas for direct, perturbative heat transport measurements. Synthetic ME-SXR data were used to optimize performance of the neural networks and study the impact of including data from various diagnostics in the networks. Initial tests on data from a previous-generation ME-SXR diagnostic on NSTX have proven successful.
| Original language | English (US) |
|---|---|
| Article number | 095015 |
| Journal | Plasma Physics and Controlled Fusion |
| Volume | 55 |
| Issue number | 9 |
| DOIs | |
| State | Published - Sep 2013 |
All Science Journal Classification (ASJC) codes
- Nuclear Energy and Engineering
- Condensed Matter Physics