Hostname: page-component-848d4c4894-ttngx Total loading time: 0 Render date: 2024-05-01T04:53:09.232Z Has data issue: false hasContentIssue false
  • English
  • Français

Statistical characterization of experimental magnetized liner inertial fusion stagnation images using deep-learning-based fuel–background segmentation

Part of: Machine Learning for Plasma Physics and Fusion Energy

Published online by Cambridge University Press:  15 September 2022

William E. Lewis Open the ORCID record for William E. Lewis [Opens in a new window] ,
Patrick F. Knapp Open the ORCID record for Patrick F. Knapp [Opens in a new window] ,
Eric C. Harding  and
Kristian Beckwith Open the ORCID record for Kristian Beckwith [Opens in a new window]
William E. Lewis*
Affiliation:
Sandia National Laboratories, Albuquerque, NM 87185, USA
Patrick F. Knapp
Affiliation:
Sandia National Laboratories, Albuquerque, NM 87185, USA
Eric C. Harding
Affiliation:
Sandia National Laboratories, Albuquerque, NM 87185, USA
Kristian Beckwith
Affiliation:
Sandia National Laboratories, Albuquerque, NM 87185, USA
*
†Email address for correspondence: willewi@sandia.gov
Get access Rights & Permissions [Opens in a new window]

Abstract

Significant variety is observed in spherical crystal x-ray imager (SCXI) data for the stagnated fuel–liner system created in Magnetized Liner Inertial Fusion (MagLIF) experiments conducted at the Sandia National Laboratories Z-facility. As a result, image analysis tasks involving, e.g., region-of-interest selection (i.e. segmentation), background subtraction and image registration have generally required tedious manual treatment leading to increased risk of irreproducibility, lack of uncertainty quantification and smaller-scale studies using only a fraction of available data. We present a convolutional neural network (CNN)-based pipeline to automate much of the image processing workflow. This tool enabled batch preprocessing of an ensemble of $N_{\text {scans}} = 139$ SCXI images across $N_{\text {exp}} = 67$ different experiments for subsequent study. The pipeline begins by segmenting images into the stagnated fuel and background using a CNN trained on synthetic images generated from a geometric model of a physical three-dimensional plasma. The resulting segmentation allows for a rules-based registration. Our approach flexibly handles rarely occurring artifacts through minimal user input and avoids the need for extensive hand labelling and augmentation of our experimental dataset that would be needed to train an end-to-end pipeline. We also fit background pixels using low-degree polynomials, and perform a statistical assessment of the background and noise properties over the entire image database. Our results provide a guide for choices made in statistical inference models using stagnation image data and can be applied in the generation of synthetic datasets with realistic choices of noise statistics and background models used for machine learning tasks in MagLIF data analysis. We anticipate that the method may be readily extended to automate other MagLIF stagnation imaging applications.

Keywords

fusion plasmaplasma diagnostics
Type
Research Article
Information
Journal of Plasma Physics , Volume 88 , Issue 5 , October 2022 , 895880501
Check for updates
Copyright
Copyright © The Author(s), 2022. Published by Cambridge University Press

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

References

REFERENCES

Anirudh, R., Thiagarajan, J.J., Bremer, P.-T. & Spears, B.K. 2020 Improved surrogates in inertial confinement fusion with manifold and cycle consistencies. Proc. Natl Acad. Sci. 117 (18), 9741–9746.CrossRefGoogle ScholarPubMed
Bishop, C. 2006 Pattern Recognition and Machine Learning. Springer.Google Scholar
Epstein, R., Goncharov, V.N., Marshall, F.J., Betti, R., Nora, R., Christopherson, A.R., Golovkin, I.E. & MacFarlane, J.J. 2015 X-ray continuum as a measure of pressure and fuel-shell mix in compressed isobaric hydrogen implosion cores. Phys. Plasmas 22 (2), 022707.Google Scholar
Ferreira, D.R., Carvalho, P.J., Fernandes, H. & JET Contributors 2018 Full-pulse tomographic reconstruction with deep neural networks. Fusion Sci. Technol. 74 (1–2), 47–56.CrossRefGoogle Scholar
Glinsky, M.E., Moore, T.W., Lewis, W.E., Weis, M.R., Jennings, C.A., Ampleford, D.J., Knapp, P.F., Harding, E.C., Gomez, M.R. & Harvey-Thompson, A.J. 2020 Quantification of MagLIF morphology using the Mallat scattering transformation. Phys. Plasmas 27 (11), 112703.CrossRefGoogle Scholar
Glorot, X. & Bengio, Y. 2010 Understanding the difficulty of training deep feedforward neural networks. Proc. Mach. Learn. Res. 9, 249–256.Google Scholar
Gomez, M.R., Slutz, S.A., Jennings, C.A., Ampleford, D.J., Weis, M.R., Myers, C.E., Yager-Elorriaga, D.A., Hahn, K.D., Hansen, S.B., Harding, E.C., et al. 2014 Experimental demonstration of fusion-relevant conditions in magnetized liner inertial fusion. Phys. Rev. Lett. 113, 155003.CrossRefGoogle ScholarPubMed
Goodfellow, I., Bengio, Y. & Courville, A. 2016 Deep Learning. MIT Press.Google Scholar
Hansen, S.B., Gomez, M.R., Sefkow, A.B., Slutz, S.A., Sinars, D.B., Hahn, K.D., Harding, E.C., Knapp, P.F., Schmit, P.F., Awe, T.J., et al. 2015 Diagnosing magnetized liner inertial fusion experiments on Z. Phys. Plasmas 22, 056313.Google Scholar
Johnson, J.M. & Khoshgoftaar, T.M. 2019 Survey on deep learning with class imbalance. J. Big Data 6 (1), 27.Google Scholar
Kingma, D.P. & Ba, J. 2017 Adam: a method for stochastic optimization. arXiv:1412.6980.Google Scholar
Kluth, G., Humbird, K.D., Spears, B.K., Peterson, J.L., Scott, H.A., Patel, M.V., Koning, J., Marinak, M., Divol, L. & Young, C.V. 2020 Deep learning for NLTE spectral opacities. Phys. Plasmas 27 (5), 052707.Google Scholar
Knapp, P.F., Glinsky, M.E., Schaeuble, M.A., Jennings, C.A., Evans, M., Gunning, J., Awe, T.J., Chandler, G.A., Geissel, M., Gomez, M.R., et al. 2022 Estimation of stagnation performance metrics in magnetized liner inertial fusion experiments using Bayesian data assimilation. Phys. Plasmas 29 (5), 052711.CrossRefGoogle Scholar
Kustowski, B., Gaffney, J.A., Spears, B.K., Anderson, G.J., Anirudh, R., Bremer, P.-T., Thiagarajan, J.J., Kruse, M.K.G. & Nora, R.C. 2022 Suppressing simulation bias in multi-modal data using transfer learning. Mach. Learn.: Sci. Technol. 3 (1), 015035.Google Scholar
Lewis, W.E., Knapp, P.F., Slutz, S.A., Schmit, P.F., Chandler, G.A., Gomez, M.R., Harvey-Thompson, A.J., Mangan, M.A., Ampleford, D.J. & Beckwith, K. 2021 Deep-learning-enabled Bayesian inference of fuel magnetization in magnetized liner inertial fusion. Phys. Plasmas 28 (9), 092701.CrossRefGoogle Scholar
Ma, T., Patel, P.K., Izumi, N., Springer, P.T., Key, M.H., Atherton, L.J., Benedetti, L.R., Bradley, D.K., Callahan, D.A., Celliers, P.M., et al. 2013 Onset of hydrodynamic mix in high-velocity, highly compressed inertial confinement fusion implosions. Phys. Rev. Lett. 111, 085004.CrossRefGoogle ScholarPubMed
McPherson, L.A., Ampleford, D.J., Coverdale, C.A., Argo, J.W., Owen, A.C. & Jaramillo, D.M. 2016 High energy x-ray pinhole imaging at the Z facility. Rev. Sci. Instrum. 87 (6), 063502.CrossRefGoogle ScholarPubMed
Rambo, P.K., Smith, I.C., Porter, J.L., Hurst, M.J., Speas, C.S., Adams, R.G., Garcia, A.J., Dawson, E., Thurston, B.D., Wakefield, C., et al. 2005 Z-beamlet: a multikilojoule, terawatt-class laser system. Appl. Opt. 44, 2421–2430.CrossRefGoogle ScholarPubMed
Rambo, P., Schwarz, J., Schollmeier, M., Geissel, M., Smith, I., Kimmel, M., Speas, C., Shores, J., Armstrong, D., Bellum, J., et al. 2016 Sandia's Z-backlighter laser facility. Proc. SPIE 10014, 1–16.Google Scholar
Rose, D.V., Welch, D.R., Madrid, E.A., Miller, C.L., Clark, R.E., Stygar, W.A., Savage, M.E., Rochau, G.A., Bailey, J.E., Nash, T.J., et al. 2010 Three-dimensional electromagnetic model of the pulsed-power $z$-pinch accelerator. Phys. Rev. ST Accel. Beams 13 (1), 010402.Google Scholar
Rovang, D.C., Lamppa, D.C., Cuneo, M.E., Owen, A.C., McKenney, J., Johnson, D.W., Radovich, S., Kaye, R.J., McBride, R.D., Alexander, C.S., et al. 2014 Pulsed-coil magnet systems for applying uniform 10–30 T fields to centimeter-scale targets on Sandia's Z facility. Rev. Sci. Instrum. 85 (12), 124701.CrossRefGoogle ScholarPubMed
Savage, M.E., LeChien, K.R., Lopez, M.R., Stoltzfus, B.S., Stygar, W.A., Artery, D.S., Lott, J.A. & Corcoran, P.A. 2011 Status of the Z pulsed power driver. In 2011 IEEE Pulsed Power Conference (eds. Curry, R.D. & Oliver, B.V.), pp. 986–990. IEEE.Google Scholar
Simpson, R.A., Mariscal, D., Williams, G.J., Scott, G.G., Grace, E. & Ma, T. 2021 Development of a deep learning based automated data analysis for step-filter x-ray spectrometers in support of high-repetition rate short-pulse laser-driven acceleration experiments. Rev. Sci. Instrum. 92 (7), 075101.CrossRefGoogle ScholarPubMed
Slutz, S.A., Herrmann, M.C., Vesey, R.A., Sefkow, A.B., Sinars, D.B., Rovang, D.C., Peterson, K.J. & Cuneo, M.E. 2010 Pulsed-power-driven cylindrical liner implosions of laser preheated fuel magnetized with an axial field. Phys. Plasmas 17 (5), 056303.CrossRefGoogle Scholar
Thiagarajan, J.J., Venkatesh, B., Anirudh, R., Bremer, P.-T., Gaffney, J., Anderson, G. & Spears, B. 2020 Designing accurate emulators for scientific processes using calibration-driven deep models. Nat. Commun. 11 (1), 5622.Google ScholarPubMed