Quantitative And Qualitative Analysis Of Olive And Flaxseed Blend Oil Based On Near Infrared Spectroscopy

Edible blend oil refers to a kind of high quality namely high price oil and one or more kinds of cheap oil,according to a certain proportion of mixing and blending,the price of blend oil is proportional to the content of high price oil.Among the products sold by illegal businesses,the content of high price oil is inconsistent with the content of high price oil marked on the bottle,or the content of high price oil accounts for only 2% and below,and still continues to produce and sell the name of high price oil as a gimmick.Based on nearinfrared spectroscopy and stoichiometry,this study used the quantitative detection and qualitative identification of two-component olive and flaxseed blend oil,three-component olive and flaxseed blend oil and the qualitative identification of multi-component and multi-type blend oil.The main contents are as follows:(1)Study and analyze the quantitative detection model and qualitative identification model of the two-component olive blend oil and the quantitative detection model and qualitative identification model of the two-component flaxseed blend oil.The optimal SVC modeling path of two components olive blend oil was AIRPLS-MSC-SG-BIPLS-GS-SVC,and the accuracy of prediction set was up to 100%.The accuracy of prediction set was up to 94.53% and the MSE was 0.11.The optimal SVC modeling path was SNV-BIPLS-GSSVC,with a prediction set accuracy of 96.43%,and the optimal SVR modeling path was SNV-CARS-GS-SVR,with a prediction set accuracy of 98.15% and a MSE of 0.16.(2)The quantitative detection model and qualitative identification model of threecomponent olive blend oil and the quantitative detection model and qualitative identification model of three-component flaxseed blend oil were studied and analyzed.The optimal SVC modeling path of three-component olive blend oil was MSC-UVE-GA-SVC,and the accuracy of prediction set was 100%.The optimal SVR modeling path was AIRPLS-MSC-SG-UVE-GA-SVR,and the accuracy of prediction set was 97.12%,and the MSE was 0.17.The optimal SVC modeling path was AIRPLS-SNV-SG-UVE-GA-SVC,and the accuracy of prediction set was as high as 100%.The accuracy of prediction set was as high as 96.52%,and the MSE was 0.36.(3)Qualitative identification models of different types of multi-component blend oils were studied and analyzed.Qualitative identification models of multi-component olive and peanut blend oils,multi-component flaxseed and peanut blend oils and multi-component olive and flaxseed blend oils were established respectively.Air PLS-UVE-GS-SVC was the best modeling path for multi-component olive and flaxseed blend oil,and the accuracy of prediction set was as high as 99.23%.The optimal modeling path of multi-component olive and peanut blend oil was AIRPLS-SNV-SG-UVE-PSO-SVC,and the accuracy of prediction set was as high as 92.47%.SNV-SG-GS-SVC was the best modeling path for the multi-component flaxseed and peanut blend oil,and the accuracy of the prediction set was as high as 97.83%.The experimental results show that,among the three kinds of parameter optimization,the GS parameter optimization modeling speed is the fastest,the(C,g)parameter obtained by GA parameter optimization is the most stable,and the established model generalization is relatively better,while the PSO parameter optimization(C,g)parameter stability is poor,but the qualitative identification model established in a specific situation has the highest accuracy.