Rapid Detection Of Adulteration Milk And Freshness Of Raw Milk By Near Infrared Spectroscopy

As a food with rich nutrient, milk and milk products has been consumpt more and more by people. But there exist some new quality problems of raw milk, such as adulteration raw milk, deterioration raw milk. They are serious threat to the safety of China's dairy industry. So it is very important to develop a quick and accurate technology to rapid detection of adulteration milk, freshness of raw milk for market management and ensure milk safety.At first, 457 samples of raw milk and adulteration milk were collected according to the adulteration concentration, these adulteration milk has seven categories which were mixed with reconstituted milk, plant cream, goat's milk, sugar, vegetable protein, starch and salt, separately, we use these raw milk and adulteration milk as study object to carried out the study of rapid detecting of adulteration milk through near-infrared (NIR) spectroscopy combined with chemometrics methods, the aim of this research was to establish NIR discriminant model between raw milk and seven kinds of adulteration milk, NIR classification discriminant model of seven kinds of adulteration milk and quantitative models of the content of adulteration material in adulteration milk.Then, non-destructive freshness assessment of raw milk during eight days of storage at 4°C was carried out by means of an FT-NIR spectrometer and a fiber optic probe. Diffuse reflectance spectra were acquired in the spectral range 12000~4000 cm-1 on samples of raw milk collected from one farm. After each spectral acquisition, the freshness parameters such as acidity, pH value and lactose content were destructively measured. For all milk samples, PLS (Partial Least Square), MLR (Multiple Linear Regression), and ANN (Artificial Neural Networks), were carried out in order to set up models to predict the freshness parameters and to classify milk samples according to the days of storage. Hierarchical cluster analysis was also conducted to test similarity between values at different days of storage.Specific results were as follows:A near infrared two categories discriminant model between raw milk and seven kinds of adulterated milk has been build,result indicated the correct distinguishing rate of 20 unknown test samples is 95%, only one sample was misjudgement. Near-infrared classification discriminant models of seven kinds of adulterated milk were then build, results indicated that classification discriminant models build among 12000~4000 cm-1 are better than models build among 7250~4250 cm-1, classification discriminant models build by MLP (Multilayer Perceptron) neural network are excelled than models build by Fisher discriminant analysis and RBF(Radial Basis Function) neural network. Finally, quantitative analysis models of the content of adulteration material in adulteration milk were build by PLS, the R2 (The Coefficient of Determination) value of quantitative calibration model of milk adulterated with reconstituted milk, milk adulterated with plant cream, milk adulterated with goat's milk, milk adulterated with sugar, milk adulterated with vegetable protein, milk adulterated with starch , milk adulterated with salt are 99.32%, 98.82%, 99.61%, 97.60%, 99.76%, 94.99%, 99.54%, separately, RMSECV (Root Mean Square Error of Calibration) are 2.58%, 0.335%, 1.93%, 0.245%, 0.0312%, 0.239%, 0.118%, separately; the R2 value of validation set are 99.36%,97.71%, 99.84%, 96.19%, 99.97%, 96.87%, 99.74%, separately, RMSEP (Root Mean Square Error of Prediction) are 2.62%, 0.478%, 1.3%, 0.309%, 0.0116%, 0.172%, 0.0867%, separately.ANN (Artificial Neural Networks) were carried out in order to set up models to predict the freshness parameters (Acidity, pH and Lactose content) of raw milk, result indicated that R2 value of quantitative calibration model are 97.09%, 96.45%, 88.22%, separately, RMSECV are 0.380 oT, 0.0202, 0.238%, separately; the R2 value of validation set are 96.47%, 98.76%, 87.72%, separately, RMSEP are 0.355 oT, 0.0256, 0.205%, separately.All of these suggested that near infrared spectroscopy has good potential to quantitative and qualitative detect raw milk rapidly and non-destructively.