| Περίληψη: | In this study, Laser Induced Breakdown Spectroscopy (LIBS) assisted with machine learning algorithms was used for the classification of milk samples based on their animal origin (i.e., cow, goat, and sheep milk). The milk samples were studied in both liquid and solid form (the solid milk samples were obtained in powdered form via lyophilization of liquid milk). Plasma formation in liquid milk samples was studied in three ways, i.e., by focusing the laser beam on the surface of the liquid sample, on sprayed milk and on the surface of a thin filament milk flow. For these four different ways of sample handling, suitable experimental setups were constructed, calibrated and the conditions of plasma generation and LIBS spectra recording were investigated. From the comparative study of LIBS spectra obtained, it was found that the configurations for the liquid and the lyophilized milk sample were the most suitable for further experimental and classification purposes. The LIBS spectra were then analyzed using machine learning algorithms.
The two algorithms applied were Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA). The former was used to reduce the dimensionality of the spectra, while the latter was used to classify/discriminate the spectra based on their animal origin. Moreover, the LDA algorithm was used to construct predictive models for evaluation of the algorithm’s effectiveness. The results obtained are very impressive and demonstrate the effectiveness of the LIBS technique assisted by machine learning algorithms for the classification of milk samples based on their animal, suggesting it as a very promising technique for applications in food safety and control in general.
|
|---|
