The Research On Near-infrared Spectroscopy Of Multimaterial Fractions Of Litsea Cubeba Fruits Calibration Model Construction

Listea cubeba(Lour)Pers is one of the unique woody oil economic trees in China with high economic value,which has great potential production application value.However,the essential oil of the Litsea cubeba fruits is volatile and the fruit is prone to deterioration,so it is particularly important for the processing enterprises to assess the quality of the fresh fruit accurately and quickly at the site of raw material acquisition.But the methods in the national standard and Chinese Pharmacopoeia,although highly accurate,are not suitable for on-site testing,and there is an urgent need to carry out research on accurate and rapid testing methods for raw material quality.This paper studied the calibration models of essential oil content,water,kernel oil and protein content of Litsea cubeba fruits based on near-infrared spectroscopy,aiming to provide a new rapid and non-destructive detection method for rapid and accurate evaluation of Litsea cubeba fruits’ quality.The main research results are as follows:123 Litsea cubeba fruit samples were collected,and the near infrared spectrum of Litsea cubeba fruits and Litsea cubeba kernels were collected by near infrared component analyzer.The essential oil,water,kernel oil and protein content of Litsea cubeba fruits were determined.The Unscrambler X10.3 software,matlab and python software were used for data analysis.The best pretreatment method was selected.The correction set and validation set were sorted by the SPXY algorithm and Kennard-stone algorithm.The redundant interference variables were eliminated by the SPA algorithm.The calibration model was established by PLS.Finally,the determination coefficient and root mean square error of the external validation set were used to evaluate theprediction performance of the established calibration model.(1)Calibration models of essential oil and water content of Litsea cubeba fruits was established.The chemical values of essential oil and water content of Litsea cubeba fruits were detected by traditional physical and chemical detection methods.The results showed that the distribution ranges of essential oil and moisture content of Litsea cubeba fruits were 2.0010%-6.9075%and 52.7581%-73.1153%,respectively.The distribution of the chemical values of essential oil and moisture were normally distributed,indicating that the samples were representative and met the conditions for establishing the calibration model.The optimal preprocessing methods for the original spectrum of essential oil and water of Litsea cubeba fruits was the second derivative combined with Detrend and SNV,respectively.the determination coefficients of the calibration models for the content of essential oil and water were all above 90%,indicating that the accuracy of the established models are good and can be used for the rapid detection of the quality of Litsea cubeba fruits.The R2 of the external validation set was 0.9086 and 0.9038 and the root mean square error was 0.3416 and 1.0621,respectively.(2)Calibration models was established for the kernel oil and protein contents of the kernels of Litsea cubeba fruits.The chemical values of the kernel oil and protein of Litsea cubeba fruits were detected by physical and chemical detection methods.The results showed that the distribution ranges of the kernel oil and protein content of the kernels of Litsea cubeba fruits were 16.967-51.9296%and 7.8700-16.9000%,respectively.The distribution of the chemical values of kernel oil and protein were normally distributed,indicating that the samples were representative and met the conditions for establishing the calibration model.The optimal pretreatment methods for the original spectrum of kernel oil and protein of the kernels of Litsea cubeba fruits were derivative SG combined with MC and second derivative,respectively.The determination coefficients of the calibration models for the content of kernel oil and protein were all above 90%,indicating that the accuracy of the established models are good and can be used for the rapid detection of kernels of Litsea cubeba fruits quality.The R2 of the external validation set was 0.9664 and 0.9110,respectively,and the root mean square error was 1.8798 and 0.4490,respectively.(3)The established models for essential oil,water,kernel oil and protein content calibrations were externally validated.The results showed that four of the 34 externally validated samples of Litsea cubeba fruits for the essential oil content calibration model had chemical values |RE|>10%.The 34 externally validated samples for the kernel oil and protein content calibration models and 32 externally validated samples for the water content calibration model all had validation results within 10%error range.The external validation values all met the requirements,indicating that the established calibration models for the multi-material components of Litsea cubeba fruits have good predictive ability.The results of this study have reference for the modeling method of complex composition samples which contains oil,and also provide a new reference method for the detection the content of essential oil,kernel oil and protein content of Litsea cubeba fruits,At the same time,it has positive significance for the development of Litsea cubeba industry.