Research On Detection Of Milk Powder Additives Based On THz Spectrum Technology

The abuse of food additives can cause serious damage to human health.THz spectroscopy is a new kind of detection technology and its wavelength is between infrared and electromagnetic wave.THz has its unique advantages.Taking melamine in milk powder and benzyl acid in flour as the research objects,the THz spectrum information was obtained,and the extracted THz optical parameters such as absorption coefficient,extinction coefficient,and refractive index were modeled to verify the feasibility of THz spectrum detection technology and good efficacy has been achieved.In addition,the samples with bags were tested and a good detection model was finally established.The main research results and conclusions are as follows:(1)First of all,this paper optimizes the process of sample preparation,and discusses the influence of different sample tablets' thickness on THz spectrum detection.It is concluded that the sample tablets' thickness about 1 mm gets the best detection results.For the samples which are difficult to tablet,adding high density polyethylene can be applied and it has very small interference to the sample.Taking milk powder,flour,arrowroot flour and additive melamine,benzoic acid,sodium cyclamate,etc as the research object and comparing the optical parameters of terahertz of the pure,this paper analyses the interrelation of the THz spectrum of the samples and their components and discuss the feasibility of THz spectrum to their detection.(2)This paper takes milk powder samples mixed with melamine as research object and explores different modeling effect under different terahertz optical parameters,modeling by using three different parameters: absorption coefficient,extinction coefficient and refractive index parameters.Firstly,Select terahertz band by using continuous projection algorithm(SPA)and Uninformative Variable Elimination(UVE),secondly,construct the partial least squares(PLS)and least squares support vector machines(LS-SVM)model.Finally,compare the strength and weakness of the six different kinds of models.The experimental results show that the three parameters all have relatively good modeling efficacy,however,the UVE-LSSVM model by adopting absorption coefficient is the best,of which the R2 and RMSEP are 0.996 and 0.5% respectively.(3)This paper uses the transmission-type detection of terahertz spectrum to explore the detection method with the packed sample.Firstly,measure 6 different kinds of polymer terahertz optical properties.This paper also takes milk powder samples mixed with melamine as research object to explore the feasibility of the detection of food additive with the packed samples.Choosing PE and PP polymer packaging materials,and the thickness is 1 mm and 3 mm respectively,the experiment had verified the feasibility of terahertz detection of samples with packing and the SPA-LS-SVM model is of the best efficacy.For the samples with PE packaging materials,after smooth algorithm pretreatment,the SPA-LS-SVM is established to predict the results and the root mean square prediction error is 0.80%;samples with PP polymer packing material,after corrected with Multiple Scatter Correction(MSC)and Normalization algorithm,the SPA-LS-SVM is established to predict the results and the root mean square prediction error is 1.05%.The experimental results show that the use of THz spectrum detection technology for the detection of food additives and food detection with packing is feasible,the research provides a theoretical guidance and technical support for the safety testing of food packing with bags.