Studies On NTR Profile And Authentication Model For Milk Powder

Milk powder is not only consumed directly,but also applied as raw material in other dairy products,milk containing drinks and foods,and highly vulnerable to adulteration and fraud,there is also in-authenticate geographical origin problem.It is necessary to grasp NIR profile and establish authentication and tracing model for milk powder.A total of 60 representative samples were collected including domestic and imported cow whole milk powder,cow sweet whole milk powder,imported skimmed milk powder,whole goat milk powder,whole yak milk powder,and non dairy cream powder,as well as dextrin,whey powder and soy protein isolate etc.food industry materials.And 480 samples adulterating the four food industry materials in the animal milk powder were prepared,and diffuse reflection spectrum were scanned simultaneously by a desktop Fourier transform NIR spectrometer and a micro NIR spectrometer,and the NIR spectrum transform,clustering and classification analysis were conducted,qualitative and semi-quantitative authentication model for milk powder were tried to establish.As results,NIR profile of milk powder and four adulterating food industrial substances are significantly different,and PCA clustering of milk powder and each adulterant shows very good separation;clustering of cow,goat and yak milk powder samples indicate isolation and separation trend,scatters of goat and yak samples clustering closer;the distance between clusters of domestic and imported cow milk powder is very far;cow skimmed milk powder,sweet milk powder and whole milk powder also all separable.It implies NIR profile can be applied to establish milk powder authentication model.To establish SIMCA milk powder authentication model,3 animals whole milk powder were adulterated with non diary cream powder,dextrin,whey powder and soy protein isolate according to 0%(pure milk powder),1%,2%,5%,10%,20%,50%and 100%(pure adulterant)percentage respectively.3D projection plots indicate that 0%,1%,and 2%adulterated sample clusters have a sense of hierarchy and separate trend,but the cluster clouds overlap between nearing groups,2%adulteration samples obviously separate for pure milk powder groups,and above 5%adulteration samples can be separated definitely and obtain reliable discrimination results.To PLS quantitative discrimination model the parameters are as follows,the maximum algorithm error arc usually within±10%,do not exceed±15%;model relative coefficient(R)above 0.9945;prediction standard error(SEP)no more than 3.3;relative standard deviation(RSD)is not more than 14.5%.It is obviously discrimination power of the portable micro NIR spectrometer is not as good as desktop Fourier transform NIR spectrometer.However,it is competent and feasible to establish discrimination model in some milk powder authentication aspect through a certain of parameters optimization.