Nuclear Engineering and Technology

Korean Nuclear Society (한국원자력학회)
권호 표지
DOI QR코드

DOI QR Code

Neutron spectrum unfolding using two architectures of convolutional neural networks

  • Maha Bouhadida (Institut de Radioprotection et de Surete Nucleaire (IRSN)) ;
  • Asmae Mazzi (Institut de Radioprotection et de Surete Nucleaire (IRSN)) ;
  • Mariya Brovchenko (Institut de Radioprotection et de Surete Nucleaire (IRSN)) ;
  • Thibaut Vinchon (Institut de Radioprotection et de Surete Nucleaire (IRSN)) ;
  • Mokhtar Z. Alaya (LMAC EA222, Universite de Technologie de Compiegne) ;
  • Wilfried Monange (Institut de Radioprotection et de Surete Nucleaire (IRSN)) ;
  • Francois Trompier (Institut de Radioprotection et de Surete Nucleaire (IRSN))
  • Received : 2022.10.20
  • Accepted : 2023.03.20
  • Published : 2023.06.25
Download PDF
⟨ Previous Next ⟩

Abstract

We deploy artificial neural networks to unfold neutron spectra from measured energy-integrated quantities. These neutron spectra represent an important parameter allowing to compute the absorbed dose and the kerma to serve radiation protection in addition to nuclear safety. The built architectures are inspired from convolutional neural networks. The first architecture is made up of residual transposed convolution's blocks while the second is a modified version of the U-net architecture. A large and balanced dataset is simulated following "realistic" physical constraints to train the architectures in an efficient way. Results show a high accuracy prediction of neutron spectra ranging from thermal up to fast spectrum. The dataset processing, the attention paid to performances' metrics and the hyper-optimization are behind the architectures' robustness.

Keywords

Acknowledgement

The authors would like to thank Stephane Gaiffas for his valuable and constructive suggestions regarding the chosen methodology. They would also like to thank Marcel Reginatto for his generous support to this research project.

References

  1. M. Reginatto, Bayesian approach for quantifying the uncertainty of neutron doses derived from spectrometric measurements, Radiat. Protect. Dosim. 121 (1) (2006) 64-69. https://doi.org/10.1093/rpd/ncl096
  2. S. Itoh, A fundamental study of neutron spectra unfolding based on the maximum likelihood method, Nucl. Instrum. Methods A 251 (1) (1986) 144-155. https://doi.org/10.1016/0168-9002(86)91161-7
  3. M. Reginatto, Overview of spectral unfolding techniques and uncertainty estimation, Radiat. Meas. 45 (10) (2010) 1323-1329. https://doi.org/10.1016/j.radmeas.2010.06.016
  4. H.N. Multhei, B. Schorr, On an iterative method for the unfolding of spectra, Nucl. Instrum. Methods A 257 (2) (1987) 371-377. https://doi.org/10.1016/0168-9002(87)90759-5
  5. C. Cao, Q. Gan, J. Song, K. Yang, L. Hu, F. Wang, T. Zhou, An adaptive deviation-resistant neutron spectrum unfolding method based on transfer learning, Nucl. Eng. Technol. 52 (2020) 2452-2459. https://doi.org/10.1016/j.net.2020.04.028
  6. S.A. Hosseini, Neutron spectrum unfolding using artificial neural network and modified least square method, Radiat. Phys. Chem. 126 (2016) 75-84. https://doi.org/10.1016/j.radphyschem.2016.05.010
  7. C. Cao, Q. Gan, J. Song, P. Long, B. Wua, Y. Wua, A two-step neutron spectrum unfolding method for fission reactors based on artificial neural network, Ann. Nucl. Energy 139 (2020), 107219.
  8. J. Wang, Z. Guo, X. Chen, Y. Zhou, Neutron spectrum unfolding based on generalized regression neural networks for neutron fluence and neutron ambient dose equivalent estimations, Appl. Radiat. Isot. 154 (2019), 108856.
  9. A.A. Alvar, M.R. Deevband, M. Ashtiyani, Neutron spectrum unfolding using radial basis function neural networks, Appl. Radiat. Isot. 129 (2017) 35-41. https://doi.org/10.1016/j.apradiso.2017.07.048
  10. R. Yamashita, M. Nishio, R.K.G. Do, Convolutional neural networks: an overview and application in radiology, Insights Imaging 9 (2018) 611-629. https://doi.org/10.1007/s13244-018-0639-9
  11. D. Im, D. Han, S. Choi, S. Kang, H.-J. Yoo, D.T.- CNN, Dilated and transposed convolution neural network accelerator for real-time image segmentation on mobile devices, in: Proc. Of the IEEE International Symposium of Circuits and Systems (ISCAS), May 26-29 2019, pp. 1-5. Saporo, Japan.
  12. O. Ronneberger, P. Fischer, T. Brox, U-Net, Convolutional networks for biomedical image segmentation, in: N. Navab, J. Hornegger, W. Wells, A. Frangi (Eds.), Medical Image Computing and Computer-Assisted Intervention - MICCAI 2015. MICCAI 2015, Lecture Notes in Computer Science, vol. 9351, Springer, Cham, 2015, https://doi.org/10.1007/978-3-319-24574-4_28.
  13. Maha Bouhadida, Mariya Brovenchenko, Vinchon Thibaut, Wilfried Monange, Francois Trompier, Neutron spectra reconstruction based on an artificial neural network trained with a large built dataset, in: ICRS 14/RPSD 2022 (14th International Conference on Radiation Shielding and 21st Topical Meeting of the Radiation Protection and Shielding Division), ANS, Seattle, United States, Sep 2022. Chal-03900950D.
  14. F. Trompier, C. Huet, R. Medioni, I. Robbes, B. Asselineau, Dosimetry of the mixed field irradiation facility CALIBAN, Radiat. Meas. 43 (2008) 1077-1080. https://doi.org/10.1016/j.radmeas.2007.11.031
  15. T. Akiba, Optuna: a next-generation hyperparameter optimization framework, in: Proc. Of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, Anchorage, USA, August 4-8, 2019, pp. 2623-2631.