Open AccessResearch on gamma spectrum semi-quantitative analysis based on convolutional neural network
Author(s) -
Fēi Li,
Jing Wang,
Liangquan Ge,
Fangyan Hu,
Cheng Feng,
K. Sun
Publication year - 2019
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1423/1/012005
Subject(s) - convolutional neural network , convolution (computer science) , computer science , artificial neural network , decomposition , field (mathematics) , nuclide , pattern recognition (psychology) , artificial intelligence , algorithm , mathematics , physics , chemistry , nuclear physics , organic chemistry , pure mathematics
In the field of practical application of nuclear technology, one of the vital steps is the interpretation of gamma energy spectrum, which can obtain the type and content of radionuclides and achieve further analyzation. Convolutional neural network is one the most basic and effective algorithm structures in deep learning because it has local receptive field, weight sharing and down sampling structure, it solves the problem of parameter expansion caused by full connection and reduces the number of weights. That can be used to study the resolution of gamma-ray spectra of convolutional neural networks. This paper introduces the spectral forming principle of gamma spectra and the convolutional neural network, and USES the convolutional neural network to study the spectral decomposition of gamma spectra. In this paper, a multi-layer convolution neural network model is built based on C# software. The convolution neural network is applied to gamma-ray spectrum decomposition, and the U, Th, K nuclides are identified and semi-quantitatively calculated. By identifying and analysing different energy spectrum data, it is shown that the modified model structure can be applied to the spectral decomposition of gamma spectrum.