https://doi.org/10.1140/epjd/s10053-023-00608-6
Regular Article – Plasma Physics
Determination of austenitic steel alloys composition using laser-induced breakdown spectroscopy (LIBS) and machine learning algorithms
Institute of Physics Belgrade, Pregrevica 118, 11080, Belgrade, Serbia
Received:
5
November
2022
Accepted:
1
February
2023
Published online:
22
February
2023
In this paper, the determination of composition of certified samples of austenitic steel alloys was done by combining laser-induced breakdown spectroscopy (LIBS) technique with machine learning algorithms. Isolation forest algorithm was applied to the MinMax scaled LIBS spectra in the spectral range form (200–500) nm to detect and eject possible outliers. Training dataset was then fitted with random forest regressor (RFR) and Gini importance criterion was used to identify the features that contribute the most to the final prediction. Optimal model parameters were found by using grid search cross-validation algorithm. This was followed by final RFR training. Results of RFR model were compared to the results obtained from linear regression with 
norm and deep neural network (DNN) by means of 
metrics and root-mean-square error. DNN showed the best predictive power, whereas random forest had good prediction results in the case of Cr, Mn and Ni, but in the case of Mo, it showed limited performance.
© The Author(s), under exclusive licence to EDP Sciences, SIF and Springer-Verlag GmbH Germany, part of Springer Nature 2023. corrected publication 2023. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
