Rapid Detection Of Quality Index Of Camellia Seed Oil By Infrared Spectroscopy

This research take the camellia oil as raw material,the use of infrared spectroscopy to study the physical and chemical index,nutrition index and fatty acid composition of rapid determination method,at the same time to quickly identify squeezing camellia oil and leaching,with different spectral data preprocessing methods and the modeling method to establish the quantitative model and identification model,to validate the model and simulation applications,based on the study of camellia oil rapid detection of basic indicators and quality.The main research contents and results are as follows:(1)A new method of rapid determination oxidant index of camellia oil was developed,This study applies the infrared spectra(IR)analysis technology combined with the method of artificial neural to predict different oxide camellia oil acid value,peroxide value and iodine value.the infrared spectra of camellia oil smoothed by SG method and pretreatmented by first derivative.Spectral data and oxidation index measured values as neural network inputs and outputs,establishing models of oxidation quantitative of camellia oil.The results show that the acid value,peroxide value and iodine value of camellia oil were measured and respectively related to the spectral data by BP neural network method,and model is used to forecast.The correlation coefficient(R)of the model of acid value,the model of peroxide value and the model of iodine value are 0.9142,0.9649 and 0.9642 respectively;Standard error(RMSE)are 0.1969 mgKOH/g,0.3926 / kg and 0.3926 gI2/100 g respectively.Simulation application of the model,acid value,peroxide value and iodine value model predicted results of the correlation coefficient(R)was 0.9819,0.9930 and 0.9819 respectively,the predict standard deviation(RMSEP)were 0.2030 mgKOH/g,0.2418 / kg and 0.6028 gI2/100 g respectively.These suggest the models of acid value,peroxide value and iodine value are reliable with good predictability and can meet the requirement of quick determination of oxidant index of camellia oil.(2)50 samples of Camellia oil from enterprise were used as materials,the infrared spectra of Camellia oil were collected and used to establish a quantitative model of the content of sterols,vitamin E and carotenoids in Camellia oil by using the method of partial least squares(PLS),and then the model was evaluated by parameters.The results showed that,the correlation coefficient of the calibration set(R)of sterol,vitamin E and carotenoid were 0.9789,0.9801 and 0.9499 respectively while the wavenumber is between 400~1850,and the root mean square error of cross validation(RMSECV)was 42.38,25.64 and 0.84mg/kg,respectively.The models were validated.The correlation coefficient of the validation set(R)of the above three components were 0.9934,0.9974 and 0.9590 respectively,the root mean square error of the validation set(RMSEP)was 13.31,6.24 and 0.18 mg/kg respectively,and the relative analysis error(RPD)were 7.769,12.693 and 2.867 respectively.Infrared spectroscopy can be used as a fast and accurate method for the determination of Camellia oil sterols,vitamin E and carotenoid content.(3)The content of oleic acid,palmitic acid and linoleic acid in Camellia oil was determined by gas chromatography,and the infrared spectrum of Camellia oil was determined by gas chromatography.Five methods used for preprocessing of infrared spectrum and then PLS with linear regression method,the SVM iPLS,siPLS and nonlinear regression method,based on the ANN infrared spectrum data with oleic acid,palmitic acid and linoleic acid content in Camellia oil quantitative model and model validation.Results show that when the oleic acid quantitative regression model is established,SG smoothing for optimal pretreatment method,ANN is adopted to establish the quantitative regression model for the optimal model of oleic acid,oleic acid of ANN model validation set and prediction set correlation coefficient R was 0.9987 and 0.9987,respectively,relative standard error less than 1%;SNV when palmitic acid quantitative regression model is established for the optimal pretreatment method,quantitative model for the optimal ANN hexadecanoic acid,hexadecanoic acid ANN model validation set and prediction set correlation coefficient R was 0.9451 and 0.9451,respectively,relative standard error less than 5%;When establishing model of quantitative analysis of linoleic acid,SD as the optimal pretreatment method,SVM and ANN linoleic acid are based on the optimal model,the model of quantitative analysis of linoleic acid of SVM model validation set and prediction set correlation coefficient of 0.9976 and 0.9742,respectively,of linoleic acid ANN model validation set and prediction set correlation coefficient was 0.9957 and 0.9957,respectively,the two models to predict relative standard error less than 1%.In conclusion,the nonlinear modeling method is more suitable for the regression model of the fatty acid content of tea oil,which indicates that the content of tea oil fatty acid is non-linear with its infrared absorption.(4)In order to standardize the market of camellia oils and safeguard the rights of consumers,establishing a rapid and accurate method to identifying pressed camellia oil and extracted camellia oil.A large number of pressed camellia oil and extracted camellia oil samples were scanned by Fourier transform infrared spectroscopy to extract the characteristic data.Savitzky-Golay smoothing(SG),multivariate scatter correction(MSC),standard normal transformation(SNV),first derivative(FD)and second derivative(SD)methods were used to preprocess,then combined with partial least squares(PLS),support vector machine(SVM)and BP neural network network,BPANN)to establish authentication model.The results showed that,When BP artificial neural network(BPANN)and partial least squares(PLS)are used to establish the discriminant models,compared with the other four kinds of preprocessing methods,SG smoothing results are the best,the correlation coefficient of validation(),the root mean square error of validation(RMSEP)and the accuracy of identification of the SG-PLS model and SG-BPANN model were 0.7679 and 0.9212,0.3226 and 0.2059,88.46% and 100% respectively.The SNV is the optimal preprocessing method for support vector machine(SVM)modeling,the correlation coefficient of validation(),the root mean square error of validation(RMSEP)and the accuracy of identification of the SNV-SVM model were 0.7614,0.8821,88.46% respectively.Therefore,the results indicating that the infrared spectroscopy can be applied to the identification of pressed camellia oils and extracted camellia oils.