Discussion On Authenticity Identification Method Of Pressing Camellia Seed Oil And Compilation Of Relevant Standard Draft

With the continuous development of the camellia seed oil(CSO)market and the continuous emergence of substandard and fake CSO products,how to solve the problem is the main problem facing our CSO market.This thesis explores the differences between pressing camellia seed oil(PCSO)and solvent extraction camellia seed oil(SECSO)in terms of basic physicochemical indicators,major components,active ingredients and volatile components.Established a standard fingerprint profile of volatile components of PCSO,a principal component analysis model and a Logistic Regression model of volatile components of PCSO and SECSO were developed respectively,which were used as the basis for preparing the standard for the"Authenticity Identification Method for PCSO".The research content and results of this paper are as follows:1.A study of the compositional differences between PCSO and SECSO.The differences of basic physicochemical indexes,composition of fatty acids and triglycerides,contents of polyphenols,tocopherols,squalene and sterols,composition and content of volatile components between PCSO and SECSO were compared.The results of the significance analysis showed that there was no significant difference in the composition and content of acid value,peroxide value,fatty acids and triglycerides between PCSO and SECSO(p>0.05);there was a significant difference in the content of polyphenols,tocopherols,squalene and sterols(p<0.05).There are also some differences in the composition and content of volatile components between PCSO and SECSO,with 2,4-dimethylhexane,Z-2-heptenal and 1-pentadecene being substances specific to SECSO.Since the stability of active components in CSO is not as good as that of volatile components,and volatile components are complex and not easy to be artificially adulterated,the volatile components of CSO were used in the identification of PCSO and SECSO.2.An investigation into the authenticity of OCSO differentiation methods based on volatile components.Three different methods were compared to distinguish PCSO from SECSO:fingerprinting,principal component analysis and machine learning models.The establishment of a principal component analysis model was not effective in distinguishing PCSO from SECSO.The standard fingerprint of volatile components of PCSO using similarity to identify PCSO from SECSO,and the similarity between PCSO samples and the established standard fingerprints is in the range of 0.805-0.929.A training set of 35 volatile CSO components was used to compare the authenticity of four machine learning models,namely logistic regression(LR),random forest(RF),support vector machine(SVM),and extreme gradient boosting(XGoost),in identifying the authenticity of PCSO,and the robustness of the four models was assessed using the leave-one-out method and triple cross-validation.The prediction results were 0.879 for LR accuracy,1 for the area under the curve(AUC),0.889 for recall and 0.97 for precision;RF,SVM and XGBoost all had the same accuracy,recall and precision of 0.876,1 and 0.876 respectively,with AUC of 0.5,0.963 and 0.806 for all three.The results of the leave-one-out method validation were 0.875 for all three models except for the logistic regression value of 0.906.The accuracy of the leave-one-out method and triple cross-validation of LR was higher than the other results,indicating that LR is suitable for the identification of the authenticity of PCSO.Using 21 CSO samples as the test set for LR,the accuracy was 90.48%,the recall was 1,and the AUC was 0.9.Therefore,a logistic regression model based on the volatile composition of CSO can effectively differentiate between PSO and SECSO.3.Notes on the preparation of the draft standard on the authenticity of PCSO.A logistic regression model based on the composition and content of volatile components in PCSO and SECSO was used as the basis for identifying the authenticity of PCSO.The draft standard for "Authenticity of Pressing Camellia Seed Oil Identification Method" was finally formed after a data search,field research of 4 CSO enterprises in Hengyang,sample collection,and data analysis.In summary,from the perspective of identification of PCSO and SECSO,this paper established a logistic regression model with volatile components as the characteristic index to predict the authenticity of PCSO sold on the market,and for the first time based on this model,compiled the standard draft of "Authenticity Identification Method of pressing camellia seed oil".This will provide a theoretical and identification method study to solve the problem that it is difficult to distinguish between PCSO and SECSO in China’s CSO market and the existence of counterfeit and shoddy phenomena,as well as to have reference significance for the quality control of other edible vegetable oils.