Study On Effect Of Temperature On Measurement Of Milk Components Contents By Near-Infrared Spectroscopy

The quality of milk is receiving the increasing attention. The contents of fat and protein in milk are the key quality index. The traditional chemistry method, time-consuming and asking for specific skill, can't satisfy the requirement of the modern quality control. It is in urgent need of developing a rapid and accurate method to determine the milk components. Our group is concentrated on the application of near-infrared spectroscopy technique for quantitative analysis of the main constituents in milk, for it is rapid,non-destructive, and environmental friendly. The paper pays specific attention on the primary study on the effect of temperature on the milk constituents'measurement by using near-infrared spectroscopy method.The optical parameters of milk can reflect milk's optical character. These parameters show the structure and biochemical character of milk from a view of optics. In Chapter 4, the effect of temperature on milk absorbance coefficient, scattering coefficient and transmission depth were studied based on the basic theory of tissue optics in combination of experiment validation, inverse added double (IAD) and Monte Carlo simulation. The following conclusions were reached, the transmission depth increased slowly with the increase of temperature. Scattering coefficient decreased gradually with the increase of temperature, and the quantitative parameter was -6.13×10^-2(mm^-1℃^-1). A complex and ruleless change was observed between absorbance coefficient and temperature, and thus the quantitative parameter can't be obtained. Further, it was deduced that the change of milk optical parameters with temperature may be caused by the change of refractive index with temperature.To compensate the effect of temperature on measurement by NIR spectroscopy, calibrations of milk constituents on temperature were developed. The method of factorial design was employed to design the levels of the factors, which helped to develop robust and accurate models. The methods of outlier detection and data pretreatment were used to reduce the effect of the other factors except for temperature. Both the global and special models on temperature improved the calibration performance. However, global model has a more excellent ability to resist the change of temperature, and the global model was more robust and adaptive.