Rapid Detection Of AFB₁ And Oxidation Process Of Edible Vegetable Oil Based On Terahertz Spectroscopy

Edible vegetable oil,as an essential consumer product of people’s daily diet,provides rich nutrients for the human body.With the frequent occurrence of edible oil safety incidents,people pay more and more attention to the quality and safety detection of edible vegetable oil.Traditional test methods of edible vegetable oil quality are tedious,time-consuming and laborious,which cannot meet the requirement of real-time monitoring.Therefore,a rapid detection method of edible vegetable oil quality and safety is urgently needed.As a new spectral detection technology in recent years,terahertz（THz）spectroscopy has lots of advantages,such as high sensitivity,strong penetration,good security and wide dynamic range.In this paper,THz spectroscopy technology combined with different chemometric methods was applied for the rapid detection of edible vegetable oil quality and safety.The main achievements of this work were summarized as follows:（1）Rapid determination of aflatoxin B₁（AFB₁）in soybean oil was studied based onTHz spectroscopy.t-distributed Stochastic Neighbor Embedding（t-SNE）and principal component analysis（PCA）,as pre-treatment methods,were used to get the best spectral features.Different chemometric methods including back propagation neural network（BPNN）,least squares-support vector machines（LS-SVM）,random forest（RF）and partial least squares（PLS）were used to obtain the best prediction model.The experimental results showed that the excellent prediction results could be obtained using BPNN combined with t-SNE,with the correlation coefficient of prediction（R_P）of 0.9948 and the root-mean-square error of prediction（RMSEP）of0.7124μg/kg.Besides,this study demonstrated that THz spectroscopy could be used to detect AFB₁at 1μg/kg in soybean oil（over 90%accuracy）.（2）THz spectroscopy was used to evaluate oxidation process of peanut oil during storage.Peroxide value（PV）of oil samples was determined according to traditional chemical methods for reference.Genetic algorithm（GA）and PCA were used as the pre-treatment methods to reduce THz high-dimensional data.Interval partial least-square（i PLS）,support vector machines（SVM）,BPNN and PLS were compared to obtain the best quantitative prediction model.The optimal absorbance spectral region related to peanut oil oxidation was found from 1.1749 to 1.5030 THz by i PLS method.In these chemometric methods used,PCA-SVM showed the best quantitative correlation between absorbance spectra and PV,with the prediction results were:R_P=0.9289 and RMSEP=1.6648 meq O₂/kg,respectively.The results of this study indicated that THz spectroscopy technology combined with appropriate chemometric methods could achieve rapid detection of mycotoxin concentrations and oxidation process in edible vegetable oil.The work provides a theoretical basis for the rapid detection of edible vegetable oil quality and safety based on THz spectroscopy,and explores a new method for the rapid detection of other food quality.