Number of the records: 1
Interpreting neural networks trained to predict plasma temperature from optical emission spectra
- 1.0584861 - ÚFCH JH 2025 RIV GB eng J - Journal Article
Képeš, E. - Saeidfirozeh, Homa - Laitl, Vojtěch - Vrábel, J. - Kubelík, Petr - Pořízka, P. - Ferus, Martin - Kaiser, J.
Interpreting neural networks trained to predict plasma temperature from optical emission spectra.
Journal of Analytical Atomic Spectrometry. Roč. 39, č. 4 (2024), s. 1160-1174. ISSN 0267-9477. E-ISSN 1364-5544
R&D Projects: GA TA ČR(CZ) TN02000009; GA ČR GF23-05186K
Institutional support: RVO:61388955
Keywords : Electron temperature * Laser induced breakdown spectroscopy * Optical emission spectroscopy
OECD category: Physical chemistry
Impact factor: 3.4, year: 2022
Method of publishing: Limited access
https://pubs.rsc.org/en/Content/ArticleLanding/2024/JA/D3JA00363A
We explore the application of artificial neural networks (ANNs) for predicting plasma temperatures in Laser-Induced Breakdown Spectroscopy (LIBS) analysis. Estimating plasma temperature from emission spectra is often challenging due to spectral interference and matrix effects. Traditional methods like the Boltzmann plot technique have limitations, both in applicability due to various matrix effects and in accuracy owing to the uncertainty of the underlying spectroscopic constants. Consequently, ANNs have already been successfully demonstrated as a viable alternative for plasma temperature prediction. We leverage synthetic data to isolate temperature effects from other factors and study the relationship between the LIBS spectra and temperature learnt by the ANN. We employ various post-hoc model interpretation techniques, including gradient-based methods, to verify that ANNs learn meaningful spectroscopic features for temperature prediction. Our findings demonstrate the potential of ANNs to learn complex relationships in LIBS spectra, offering a promising avenue for improved plasma temperature estimation and enhancing the overall accuracy of LIBS analysis.
Permanent Link: https://hdl.handle.net/11104/0352649
File Download Size Commentary Version Access 0584861.pdf 0 2.2 MB Publisher’s postprint require 0584861preprint.pdf 0 2.2 MB Author´s preprint require
Number of the records: 1