Classifcation of brandy products by geographical origin and ageing based on raman spectra and discriminant analysis methods

The issues of developing simple and affordable methods for controlling the authenticity and quality of alcoholic beverages based on Raman spectroscopy are considered. The most important characteristics of brandy and cognac products – geographical origin and aging period, which largely determine the cost of production and by which it is often falsified, are studied. The advantages of the investigated method include ease of sample preparation up to its complete absence, high selectivity, rapidity and simplicity of analysis, the possibility of developing compact instruments that allow analysis to be carried out directly on the sampling spot. The Raman spectra of 42 different samples of brandy and cognac products, differing in geographical origin and aging period, were measured. It is shown that the fragments of the spectra measured in the range of Raman shifts from 800 cm^–1 to 3000 cm^–1 are the most informative for solving the tasks set. From the studied samples, training and test sets were formed. For data processing, machine learning models trained using the extreme gradient boosting algorithm were used. The correctness of recognition by geographical origin and aging period for undiluted samples of the test set, the spectra of which were not used when training the model, was 100 %. The results of the study can be used to develope compact devices for express control of the authenticity of alcoholic products and determine their characteristics using Raman spectra and their further processing by machine learning methods.

1. Cantarelli M. Á., Azcarate S. M., Savio M., Marchevsky E. J. Food Analytical Methods, 2015, vol. 8, no. 3, pp. 790–798. https://doi.org/10.1007/s12161-014-9958-8

2. Bozkurt S. S., Merdivan E., Benibol Y. Microchimica Acta, 2010, vol. 168, no. 1, pp. 141–145. https://doi.org/10.1007/s00604-009-0271-y

3. Nordon A., Mills A., Burn R. T., Cusick F. M., Littlejohn D. Analytica Chimica Acta, 2005, vol. 548, no. 1-2, pp. 148–158. https://doi.org/10.1016/j.aca.2005.05.067

4. Ellis D. I., Eccles R., Xu Y., Griffen J., Muhamadali H., Matousek P., Goodall I., Goodacre, R. Scientific reports, 2017, vol. 7, 12082. https://doi.org/10.1038/s41598-017-12263-0

5. Houhou R., Bocklitz T. Analytical Science Advances, 2021, vol. 2, no. 3-4, pp. 128–141. https://doi.org/10.1002/ansa.202000162

6. Berghian-Grosan C., Magdas D. A. Scientific Reports, 2020, vol. 10, 21152. https://doi.org/10.1038/s41598-020-78159-8

7. Deneva V., Bakardzhiyski I., Bambalov K., Antonova D., Tsobanova D., Bambalov V., Cozzolino D., Antonov L. Molecules, 2019, vol. 25, no. 1, 170. https://doi.org/10.3390/molecules25010170

8. Ranaweera R. K., Gilmore A. M., Capone D. L., Bastian S. E., Jeffery D. W. Food Chemistry, 2021, vol. 335, 127592. https://doi.org/10.1016/j.foodchem.2020.127592

9. Natekin A., Knoll A. Frontiers in Neurorobotics, 2013, vol. 7, 21. https://doi.org/10.3389/fnbot.2013.00021