Effect Of Temperature And Homogenization On Near-Infrared Spectroscopy/Dielectric Spectroscopy And Quality Prediction Accuracy Of Milk

Raw milk is rich in nutrition and the rapid and accurate detection of its components is directly related to the quality evaluation of milk and the healthy development of dairy industry.Near-infrared spectroscopy and dielectric spectroscopy technology,as modern detection technology,have the advantages of fast detection,low-cost detection and nondestructive detection.They have been widely used in milk quality detection,but in the detection process of electromagnetic spectrum method,it is easy to be affected by the heterogeneous characteristics of milk,and has the problem of poor measurement repeatability,which further affects the detection accuracy.Therefore,it is of great significance to explore the effect of improving the repeatability of spectral measurement on the accuracy of the prediction models of milk major components contents based on near-infrared spectroscopy and dielectric spectroscopy.In this paper,the effects of temperature and homogenization treatment on near-infrared spectroscopy and dielectric spectroscopy of milk were analyzed by using measurement repeatability error as evaluation method.The prediction models of main components contents in milk were established under different treatment conditions(normal temperature,suitable measurement temperature and suitable homogenization pressure),and the accuracy of prediction model was compared and analyzed.The main research contents and conclusions are described as follows:(1)The effects of temperature on milk absorbance,relative dielectric constant(?')and dielectric loss factor(?")were studied.The results showed that the absorbance of milk changed with temperature in different bands.For the dielectric spectroscopy,?'decreased with increasing temperature above 60 MHz.?"increased with temperature below 1000 MHz.Above 2000 MHz,?"decreased with increasing temperature.(2)The effects of temperature on the repeatability error of milk absorbance(?_A),the repeatability error of relative dielectric constant(?_?')and the repeatability error of dielectric loss factor(?_?")were studied.The results show that the value of?_A at long wavelengths was greater than that at short wavelengths for NIR.The?_A increased with temperature from25°C to 35°C.The?_A decreased with increasing temperature between 35°C and 40°C,and reached the minimum at 40°C.For the dielectric spectroscopy,?_?'and?_?"increased with increasing temperature between 25°C and 30°C.The?_?'and?_?"decreased with increasing temperature between 30°C and 40°C,and reached the minimum at 40°C.The optimum temperature was determined to be 40°C.(3)The effects of homogenization on milk absorbance,?'and?"were studied.The results showed that the absorbance of milk was significantly reduced by homogenization for NIR.However,between 20/5 MPa and 90/5 MPa,there was little difference in the absorbance of homogenized milk under different pressures.For the dielectric spectroscopy,homogenization increased the values of?'and?",but there was little difference between?'and?"under different pressures.The effects of homogenization on the?'and?"of milk were summarized as follows:zeta potential on the surface of fat globules,viscosity and hydration of milk,gravity emulsification and particle aggregation.(4)The effects of homogenization on milk?_A,?_?'and?_?"were studied.The results showed that the value of?_A was significantly reduced by homogenization treatment at the appropriate pressure,which indicated that homogenization treatment at the appropriate pressure improved the measurement repeatability of NIR of milk.The?_A at 70/5 MPa was the smallest in the full wavelength range.For the dielectric spectroscopy,the?_?'and?_?"of milk were significantly reduced by homogenization treatment.(5)The effects of temperature and homogenization on the prediction accuracy of milk major components based on NIR were studied.The results showed that improving the repeatability of spectral measurement by changing the temperature and homogenization treatment could improve the prediction accuracy of milk fat and water content.Before homogenization,the maximum residual prediction deviations of milk fat and water content based on NIR were 2.4603 and 2.7466.After homogenization,the maximum residual prediction deviations were 4.0507 and 3.8053.(6)The effects of temperature and homogenization on the prediction accuracy of main components of milk based on the dielectric spectroscopy were studied.The results showed that improving the repeatability of spectral measurement by changing the temperature and homogenization treatment could improve the prediction accuracy of milk fat and water content.Before homogenization,the maximum residual prediction deviations for predicting milk fat and water content based on the dielectric spectroscopy were 1.8187 and 1.7564.After homogenization,the maximum residual prediction deviations were 2.2943 and 2.6698.Comparing the prediction accuracy,it was found that homogenization had a greater effect on the prediction accuracy of milk fat and water content based on NIR.