Rapid evaluation of milk acidity and identification of milk adulteration by Raman spectroscopy combined with chemometrics analysis

Milk can provide people with rich nutrition, and its freshness is widely concerned. Acidity, as an important indicator of freshness, can evaluate the freshness of milk. In this paper, a new method for rapid and non-destructive evaluation of milk acidity based on Raman spectroscopy and chemometrics is proposed. Two multivariate data analysis methods, partial least square regression (PLSR) and back propagation artificial neural network (BP-ANN), are used to build models. The determination coefficient (R2), root mean square error (RMSE) and the ratio of prediction to deviation (RPD) values for the best prediction model BP-ANN test sets are 0.9724, 0.7223 degrees T and 6.31, respectively. The results show that nonlinear model BP-ANN is more suitable for predicting the acidity of milk. Furthermore, the model can identify the illegal neutralization of the acidity of milk by adding Na2CO3. Thus, Raman spectroscopy with chemometrics is a fast and powerful tool to assess milk acidity. Finally, we conclude with an in-depth discussion and analysis of the model results to further validate that the acidity is predicted by establishing a link with the various components of milk such as protein and fat.

Automated summary

This study proposes a rapid and non-destructive method for evaluating milk acidity based on Raman spectroscopy and chemometrics. It utilizes nonlinear modeling to predict the acidity and can detect illegal neutralization of milk acidity.