Study of the possibility of using an artificial neural network to recognize and assess the contribution of individual radionuclides to the total beta spectrum
The aim of this work is to study the possibility of using an artificial neural network for identification and quantitative assessment of the content of individual radionuclides in the total beta spectrum. The neural network implemented by using of Matlab R2020b. A single-layer feedforward neural net...
Saved in:
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Saint-Petersburg Research Institute of Radiation Hygiene after Professor P.V. Ramzaev
2020-12-01
|
| Series: | Радиационная гигиена |
| Subjects: | |
| Online Access: | https://www.radhyg.ru/jour/article/view/755 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849340691445972992 |
|---|---|
| author | V. S. Repin |
| author_facet | V. S. Repin |
| author_sort | V. S. Repin |
| collection | DOAJ |
| description | The aim of this work is to study the possibility of using an artificial neural network for identification and quantitative assessment of the content of individual radionuclides in the total beta spectrum. The neural network implemented by using of Matlab R2020b. A single-layer feedforward neural network with one invisible layer of 10 neurons and 3 outputs (according to the number of radionuclides) was used. To test and study the capabilities of the artificial neural network, 3 smooth model spectra were selected — 40K, 137Cs and 90Sr, obtained on the liquid spectrometer Quantulus 1220. The results of the study showed that neural networks are an effective method for recognizing of the contribution of an individual radionuclide or establishing its presence in the total beta spectrum. The recognition accuracy depends on the smoothness of the spectrum and does not exceed 30% if the share of the radionuclide in the total spectrum is more than 10%, which is quite suitable for practical use. For statistically «noising» spectra, the method can be used to preliminary estimate the weight coefficients of individual radionuclides, the final value of which can be obtained by minimization methods with subsequent statistical criterial fitting of the total spectrum shape. |
| format | Article |
| id | doaj-art-83f7ec7e0f5a4b1d915501f3a66297a4 |
| institution | Kabale University |
| issn | 1998-426X |
| language | English |
| publishDate | 2020-12-01 |
| publisher | Saint-Petersburg Research Institute of Radiation Hygiene after Professor P.V. Ramzaev |
| record_format | Article |
| series | Радиационная гигиена |
| spelling | doaj-art-83f7ec7e0f5a4b1d915501f3a66297a42025-08-20T03:43:51ZengSaint-Petersburg Research Institute of Radiation Hygiene after Professor P.V. RamzaevРадиационная гигиена1998-426X2020-12-01134748110.21514/1998-426X-2020-13-4-74-81676Study of the possibility of using an artificial neural network to recognize and assess the contribution of individual radionuclides to the total beta spectrumV. S. Repin0Saint-Petersburg Research Institute of Radiation Hygiene after Professor P.V. Ramzaev, Federal Service for Surveillance on Consumer Rights Protection and Human Well-BeingThe aim of this work is to study the possibility of using an artificial neural network for identification and quantitative assessment of the content of individual radionuclides in the total beta spectrum. The neural network implemented by using of Matlab R2020b. A single-layer feedforward neural network with one invisible layer of 10 neurons and 3 outputs (according to the number of radionuclides) was used. To test and study the capabilities of the artificial neural network, 3 smooth model spectra were selected — 40K, 137Cs and 90Sr, obtained on the liquid spectrometer Quantulus 1220. The results of the study showed that neural networks are an effective method for recognizing of the contribution of an individual radionuclide or establishing its presence in the total beta spectrum. The recognition accuracy depends on the smoothness of the spectrum and does not exceed 30% if the share of the radionuclide in the total spectrum is more than 10%, which is quite suitable for practical use. For statistically «noising» spectra, the method can be used to preliminary estimate the weight coefficients of individual radionuclides, the final value of which can be obtained by minimization methods with subsequent statistical criterial fitting of the total spectrum shape.https://www.radhyg.ru/jour/article/view/755neural networksliquid beta spectrometryspectrum decompositiontraining samplereliability |
| spellingShingle | V. S. Repin Study of the possibility of using an artificial neural network to recognize and assess the contribution of individual radionuclides to the total beta spectrum Радиационная гигиена neural networks liquid beta spectrometry spectrum decomposition training sample reliability |
| title | Study of the possibility of using an artificial neural network to recognize and assess the contribution of individual radionuclides to the total beta spectrum |
| title_full | Study of the possibility of using an artificial neural network to recognize and assess the contribution of individual radionuclides to the total beta spectrum |
| title_fullStr | Study of the possibility of using an artificial neural network to recognize and assess the contribution of individual radionuclides to the total beta spectrum |
| title_full_unstemmed | Study of the possibility of using an artificial neural network to recognize and assess the contribution of individual radionuclides to the total beta spectrum |
| title_short | Study of the possibility of using an artificial neural network to recognize and assess the contribution of individual radionuclides to the total beta spectrum |
| title_sort | study of the possibility of using an artificial neural network to recognize and assess the contribution of individual radionuclides to the total beta spectrum |
| topic | neural networks liquid beta spectrometry spectrum decomposition training sample reliability |
| url | https://www.radhyg.ru/jour/article/view/755 |
| work_keys_str_mv | AT vsrepin studyofthepossibilityofusinganartificialneuralnetworktorecognizeandassessthecontributionofindividualradionuclidestothetotalbetaspectrum |