Research On Multi-Spectral Fusion Technology Of Olive Oil Adulteration Identification And Olive Blended Oil Detection

Olive oil is made by cold-pressing fresh olive oil fruits without heating or other chemical treatment during the preparation process.While retaining natural nutrients,it is extremely rich in monounsaturated fatty acids.Olive oil consumption has huge consumption prospects in China.Research and investigation of the regular consumption of olive oil can degrade the body's fats and lower blood pressure,and effectively prevent cardiovascular and cerebrovascular diseases,so it has the reputation of "liquid gold".At present,there are some unscrupulous traders who mix low-value edible oil into olive oil and sell adulterated olive oil to make high profits,For the sale of blended olive oil,there is also false blending oil that does not meet the label blending ratio.Therefore,it is very necessary to analyze and study a method of adulteration of olive oil and a rapid detection method of false olive blending oil.Used near infrared,Raman and mid-infrared spectroscopy fusion technology combined with stoichiometry and machine learning to analyze the problem of olive oil adulteration identification and false olive blending oil detection,and established qualitative and quantitative identification detection of olive oil adulteration Model,qualitative and quantitative adulteration of olive oil and quantitative detection model of olive blended oil,the main research content and results are as follows:(1)Analyzed and established the qualitative identification and quantitative detection model of binary and ternary adulterated olive oil based on near-infrared and mid-infrared spectroscopy technology,studied and analyzed the optimal modeling method and technical route,and established the near-infrared binary doped The pseudo-qualitative identification of baseline-CARS-PLS-SVC model and the mid-infrared binary adulterated model of MSC-CARS-PLS-SVC were both 100% accurate in prediction set of sample bank,and the correlation coefficients of the prediction sets of the near-infrared and mid-infrared olive oil binary adulterated quantitative detection models were higher than 98%,and the mean square error MSEP was lower than 0.05,which had a good quantitative prediction effect.The near-infrared ternary olive oil adulterated MSC-PCA model had a good qualitative identification and quantitative detection effect.(2)Analyzed and established olive oil binary adulteration qualitative identification and quantitative detection model based on near infrared-middle infrared spectral data fusion and near infrared-raman spectral data fusion and middle-raman spectral data fusion.Research showed that the quantitative detection model based on data fusion in feature level of three spectral technology had good performance,having higher recognition accuracy rate and a small prediction deviation.Studies had shown that the feature layer fusion model was superior to the data layer fusion model,compared with the single light model,the fusion model had better stability and accuracy.ty and accuracy of the model.(3)Analyzed and established a reconstructed olive oil support vector machine detection model based on near infrared spectroscopy technology,in which the correlation coefficient of the DOSC-CARS-PLS-GC-SVR model was as high as 99.74%,the mean square error MSEP was 0.08,and the model has good quantitative detection accuracy.At the same time,in view of the poor prediction performance of the SG-CARS-PLS-SVR model,the near-infrared feature layer data fusion method could effectively improved the prediction accuracy of the model.