Monitoring of the main indicators of milk quality of Ukrainian dairy produc-ers

The composition and properties of milk are significantly influenced by various factors, such as the breed of cows, stages of lactation, the environment, as well as its temperature treatment. In the research carried out in 2017–2022, ultrasonic analysis of 56 samples of different types of dairy products was carried out: pasteurized and ultra-pasteurized milk of the largest Ukrainian producers of dairy products and milk produced at small enterprises, as well as a dairy product made in laboratory conditions from milk powder (reconstituted milk). The following indicators of milk composition were determined by the method of ultrasonic analysis, such as the content of fat, protein, lactose, solids non fat (SNF), and freezing temperature. The fat content in the samples was in the range of 2.5–3.2 %, the protein content was on average 3.3%, and a clear pattern of increasing the protein content with increasing fat content was observed. The content of SNF was on average 8.4 %, and lactose in the range of 4.5–5.5 g/100 g of milk; in the produced reconstituted milk, the lactose content was somewhat lower and averaged 3.5 %. The freezing temperature of the studied samples ranged from -0.50 ºC (reconstituted milk) to -0.60 ºC (pasteurized and ultra-pasteurized). The method of principal components analysis was used to process the obtained data array. Application of the method of principal components analysis (PCA) made it possible to divide all the studied samples into two groups. The first compact group consists mainly of samples of pasteurized and ultra-pasteurized milk from well-known producers of dairy products. The second diffuse group consists of samples of reconstituted milk, and this group includes several samples of pasteurized milk of a low price category from small producers. The analysis of the eigenvectors and eigenvalues of the PCA indicates that the main factors affecting the grouping of the samples are the content of lactose, protein and the value of the freezing temperature. At the same time, the application of PCA modeling did not reveal any grouping of samples by seasons and years. In the future, based on this method, a discriminatory model can be made, which will allow to quickly detect falsified samples or samples manufactured in violation of technical conditions.