[3C10-15] Dose Estimation

3:45 PM - 4:00 PM JST(6:45 AM - 7:00 AM UTC)

[3C14] Estimation of radiation source and shields distribution from measured gamma-ray energy spectra using machine learning

*Yuka Kumada¹, Taiyo Sato¹, Tsugiko Takase¹, Masaharu Matsumoto¹, Katsuhiko Yamaguchi¹ (1. Fukushima Univ.)

Keywords:

Machine Learning,Radiation source distribution estimation,PHITS,γ-ray energy spectrum

Using a radiation source distribution estimation technique using machine learning, we verified the estimation accuracy of radiation source distribution and shielding position when a shielding object is present between a radiation source and a detector arranged on a two-dimensional plane. The measured spectrum data was input into a first-stage neural network and corrected to be close to the shape of the spectrum obtained by a simple simulation calculation, and then the distribution of the radiation source and shields were input into a separate neural network for estimating radiation source and shields distribution, and the accuracy was verified.