Rapid Detection Of Trans Fatty Acid Content In Edible Vegetable Oil By Near Infrared And Raman Spectroscopy

In recent years,more and more studies on Trans Fatty Acids（TFA）show that eating into TFA may cause potential harm to human health.Edible vegetable oil is one of the most widely used edible oils in people’s daily life,with the improvement of processing technology and the appearance of more and more hydrogenated vegetable oil,the content of TFA in edible vegetable oil may be exceeded.Therefore,for the consumer’s health,it is very important to study a fast method to detect the TFA content in edible vegetable oil.Near Infrared（NIR）and Raman spectroscopy is a green,fast and nondestructive detection method.In this paper,the TFA content in edible vegetable oil was studied using NIR,Raman spectroscopy and Chemometrics method,and the model was optimized.Its main research contents are lised as follows:The content of TFA in edible vegetable oil was detected by NIR spectroscopy.The NIR spectra of edible vegetable oil samples were collected in the range of 4000¹0000 cm^-1 wave number,and the NIR model of TFA content in edible vegetable oil was established by optimizing the band selection,pretreatment method selection,variable selection and modeling method.The model results show that the coefficient of determination（R²）of the optimal prediction model in calibration set and prediction set are 0.992 and 0.989,root mean square error of calibration（RMSEC）and root mean square error of prediction（RMSEP）are0.071%and 0.075%respectively.Compared with the model results before optimization,the number of variables used in the optimal prediction model only accounted for 1.215%of the total band variables,while the R² of the optimal predictive model in predictive set rise from0.904 to 0.989,and RMSEP drop from 0.230%to 0.075%.The content of TFA in edible vegetable oil was rapidly detected by Raman spectroscopy.The Raman spectra of edible vegetable oil samples were collected in the range of279.08²301.1 cm^-1 wave number,and the Ramna spectral data were optimized by background deduction,normalized pretreatment,band selection and variable selection,then the partial least-squares regression（PLS）model of TFA content in edible vegetable oil was established.Compared with the optimal model,the coefficient of determination（R²）of calibration set and prediction set are increased from 0.9431,0.8634 to 0.9716,0.9130respectively.RMSEC and RMSEP are reduced from 0.1843%,0.2875%to 0.1302%,0.1926%respectively,and the number of variables used in the optimal model modeling only accounted for 7.03%of the variables used in the model before optimization.Therefore,the above researches show that the application of NIR and Raman spectroscopy can be used for rapid and nondestructive detection of TFA content in edible vegetable oils,and the model optimization is necessary to simplify the model and improve the accuracy and prediction ability of the model.