Modern dairy factories produce thousands of cheese blocks per day. Cheese quality is partly defined by the concentration of dry matter and fat. In this study, we evaluated three different near infrared spectroscopy instruments for on-line determination of fat and dry matter in cheese blocks of approx. size 35 × 28 × 12 cm: scanning reflection (908–1676 nm), scanning interaction (760–1040 nm), and imaging interaction measurements (760–1040 nm). The near infrared measurements were performed on fresh cheese blocks in a pilot plant at three different critical control points (CCP): (CCP1) before pressing, (CCP2) after pressing, and (CCP3) after salting. A total of 160 cheeses from 10 production batches were measured. Whereas near infrared measurements were obtained from the surface of the cheese blocks, the reference analysis was done on a cross-section of the cheese blocks. In general, good results were obtained regressing the reference values onto the near infrared measurements using partial least squares regression. For example, using near infrared scanning reflection at CCP2 yielded root mean squared errors of cross-validation on 0.44% and 0.64% for fat and dry matter, respectively. Hence, surface chemistry of cheese blocks were representative for the average chemistry of the blocks. Furthermore, this study finds that it is possible to predict fat and dry matter at CCP3 based on near infrared measurements obtained at CCP1 earlier in the process. This enables improved control of the cheese making process, as it is possible to detect deviations from target quality early in the production process.
Assessment of avocado fruit dry matter content using portable near infrared spectroscopy: Method and instrumentation optimisation
