[1N10-12] Applications of Quantum Beam and Targets 2

[1N11] Wavelength-resolved Neutron Imaging Data Analysis by Combining Slow Neutron Total Cross-section Calculation Code with Machine Learning

*Tomoki Kasahara¹, Hirotaka Sato¹, Takashi Kamiyama¹ (1. Hokkaido Univ. )

Keywords:Wavelength-resolved Neutron Imaging, Machine Learning, Bragg Edge, Crystalline Microstructure Analysis, RITS

Neutron transmission spectrum from a polycrystalline material shows the Bragg-edge profile. Extraction of crystalline microstructural information from Bragg-edge spectrum is realized by RITS which is a slow neutron transmission spectral analysis code. Spectral fitting analysis using RITS requires precise initial parameter estimation of crystalline structure, and it is difficult for non-specialists. Therefore, in this study, machine learning of various patterns of slow neutron total cross-sections calculated by RITS was used to analyze the measured spectra. The results analyzed by this new method were compared with the results analyzed by RITS. As a result, the real-space distributions of bulk crystalline microstructural parameters of steel samples obtained by the new method were consistent with ones obtained by RITS.