Research On The Authentication And Adulteration Detection Of The Edible Oil By Broadband Terahertz Spectroscopy And Chemometrics

Edible oil is an essential food in daily life and provides basic nutrients including vitamins,essential fatty acids and trace elements.The quality of edible oil has always been one of the most important issues in the field of food safety.Since edible oil is a complex food matrix,it is difficult to use a method to effectively detect all substandard products.Terahertz spectroscopy is an electromagnetic wave between infrared and microwave band.Since THz wave contains physical,chemical and structural information of material,it has a wide range of applications in physical chemistry,biological medicine,materials science and other fields.As a kind of new detection technology,THz has potential application value in the field of edible oil detection.Chemometrics,derived from mathematics,statistics and computer science,is widely used in chemical analysis,metabolomics,biology and other fields,and is also an important analysis method in spectroscopy.This article will use broadband terahertz time-domain spectroscopy combined with chemometrics for qualitative identification and quantitative analysis of edible oil.In this paper,the main research contents include:1.The spectral characteristics generated by organic crystal DSTMS as a generator to emit terahertz radiation and Fabry-Perot echo phenomenon are states.The THz time-frequency characteristics in nitrogen environment are analyzed by using Hilbert-Huang Transform and a adaptive filtering method based on empirical mode decomposition(EEMD)and the soft threshold wavelet filtering(WT)is proposed.We also analyze Hilbert spectrum and time-frequency characteristics of THz signal of three different edible oils and provide a new method of time-frequency analysis for the material detection.2.Five kinds of commonly used edible oil are measured by using terahertz time-domain spectroscopy system.Refractive index and absorption spectrum characteristics of the samples are analyzed in 1.2-6.6 THz range.A method of a immune genetic algorithm(IGA)and partial least squares discriminant analysis(PLS-DA)algorithm is proposed for classification of edible oils.The several traditional variable selection algorithm based on PLS model including interval partial least squares(i PLS),backward interval partial least squares(biPLS)and moving window partial least squares(mwPLS)are compared.The experimental results show that immune genetic algorithm combined with partial least squares discriminantanalysis algorithm has achieved the highest classification accuracy.3.The origin of extra virgin olive oil are analyzed by using terahertz time-domain spectroscopy and chemometric method.The three different origin of extra virgin olive oil samples from PDO region are chosen for research.These samples come from Baena region in Spain,Mazara region in Italy and Mylopotamos region in Greek,respectively.The refractive index and absorption spectrum of these samples are measured and analyzed in 1.4-5.0 THz.The principal component analysis(PCA)is used to extract the spectral feature variables as the input variables of support vector machine(SVM)model,and an improved adaptive chaotic particle swarm optimization algorithm(ACPSO)is used to optimize the parameter of SVM model.The methods of parameters optimization of standard particle swarm algorithm and genetic algorithm are compared.The experimental results show that the PCA-ACPSO-SVM model combined with THz absorption spectrum is a new analysis method for origin identification of olive oil and achieve better classification results,with the classification accuracy of 96.91% and 98.08% in the calibration set and prediction set,respectively.4.The terahertz spectroscopy and chemometric method are used for qualitative identification and quantitative analysis of adulterated olive oil.The olive oil samples mixed with sunflower seed oil in different concentrations(10%,30%,50%,70%,90%)are measured in the terahertz band.The refractive index and absorption spectrum of these samples are analyzed in 1.5-4.9 THz and the linear relationship between absorption spectrum intensity and the concentration of adulteration at 2.75 and 4.45 THz are analyzed.SIMCA classification method is used for qualitative identification,and principal component regression(PCR),partial least squares regression(PLSR),least squares support vector machine(LSSVM)and RBF neural network(RBFNN)are applied for quantitative analysis of adulterated concentration.The experimental results show that the SIMCA method has classification accuracy of100%,the RBFNN model has the highest prediction accuracy,with calibration correlation coefficients(R_c~2)of 0.9986,prediction correlation coefficients(R_p~2)of0.9982,the calibration root mean square error(RMSEC)of 1.2243 and prediction root mean square error(RMSEP)of 1.3644.In this paper,THz spectroscopy combined with chemometrics has been used to study the qualitative detection and quantitative analysis of edible oil.The feasibility and effectiveness of this method are verified and a new method is proposed foridentification and adulteration detection of edible oils.As a kind of fast,reliable,and nondestructive detection method,it can be applied to similar food and agricultural products detection,provides a new detecting method and important significance in the fields of food safety and quality monitoring.