S. Atanassova, N. Naydenova, T. Kolev, T. Iliev, G. Mihaylova
Abstract. . The aim of the study has investigation of the feasibility of Near infrared(NIR) spectroscopy in a diffusion reflection mode as a tool for the monitoring a chemical changes during ripening of Bulgarian yellow cheese from cow milk and classification of cheese according to ripening stage. The cheeses were made according to Bulgarian national standard for Bulgarian yellow cheese from cow milk. Samples for spectroscopic and chemical analysis were taken after 1, 5, 10, 15, 20, 25, 30, 35, 40 and 45 days of ripening. Total protein, water-soluble protein, titratable acidity, dry matter, and sodium chloride content were determined for all samples. NIR measurements were performed by NIRQuest 512 spectrometer (Ocean Optics, Inc.) in the region 900-1700 nm using reflection fiber-optics probe. Partial least square regression with internal cross-validation was used for calibration models development for determination of tested parameters. Soft Independent Modeling of Class Analogy (SIMCA) was implemented to create models of the cheese samples according to maturity stages based on their NIR spectra. Differences in spectral data of cheese samples during ripening were found. The biggest variation in spectral data were observed at 1215, 1350, 1644 nm, region from 1415 to 1470 and 1682-1700 nm. Determination of water-soluble protein, dry matter content and titratableacidity with coefficient of multiple correlation bigger than 0.95 and ratio of standard error of calibration and standard deviation of data set bigger than 3 was excellent. Very good determination was found for maturity rate. Prediction accuracy was good for total protein content and poor for NaCl content of cheese. SIMCA model for discrimination of cheese samples according to ripening stage was developed. We can conclude from obtained results that NIR has a potential for assessing cheese ripening related to changes in the cheese matrix during maturity.