Study On Maturity Evaluation Index Mobel Of Wine Grape Based On Near Infrared Spectroscopy

The sugar,acid,and polyphenols in wine grape berries are the main evaluation factors for judging the maturity of grapes,but the evaluation factors are affected by factors such as climate,soil,and variety,and their contents cannot be synchronized to the optimal level during the grapes ripening.Therefore,in order to integrate the effects of various indexes on grape berry quality,a principal component analysis(PCA)method was used to establish evaluation model of wine grapes,and the main evaluation factors were selected based on the influence of each index and the correlation between polyphenol indexes.Near infrared spectroscopy was used to establish a rapid detection method for the content of the main evaluation indexes in wine grapes based on the results of principal component analysis.In this study,Meili(ME)and Muscat Hamburg(MH)from Yang Ling area of Shaanxi province were empolyed as testing subjects.Near infrared spectroscopy data of grape varieties in different ripening periods were collected and the phyto-chemical indexes and polyphenol contents of each ripening period were determined.In order to study a rapid detection method for the content of each evaluation indexes and quality evaluation model for wine grape berries.The main conclusions of this experiment were as follows:1.The information of 14 phyto-chemical indexes and polyphenol indexes was integrated by the principal component analysis method to establish evaluation model of wine grape quality in different ripening periods.It was concluded that the best harvest time of ME and MH was Sep 1.According to the correlation between each index and the influence of each evaluation index on the evaluation model,total acid,sugar-acid ratio,total phenols in skins,total anthocyanins in skins and total phenols in seeds were extracted as the main evaluation indexes of the quality of ME grapes,soluble solids contents,total phenols in skins,total tannins in seeds,total phenols in seeds,and total tannins in seeds as the main evaluation indexes of the quality of MH grape.2.(1)The influences of different pretreatment methods and wavelength intervals on the soluble solids contents(SSC)model of ME grapes were compared.The ME SSC model established by first derivative and partial least squares regression(PLSR)in the wavelength ranges of 950～985 nm,1130～1160 nm,and 1200～1265 nm was the best,RMSEC and RMSEP was 0.6883 and 0.5811 respectively,Rcal and Rp was 0.9181 and 0.9364 respectively,and RPD was 2.8513,which had good predictive accuracy and fitting effect.After paired sample T-tests,it was no significant difference between the predictive values and the actual values in validation set,the mean values of the predicted values and the actual values measured by the traditional methods at different ripening periods.The established model had a significant predictive effect on the changes of the SSC of ME grape at different ripening periods.(2)The total acid(TA)model established in 950～985 nm,1165～1195 nm,and1340～1405 nm wavelength ranges after SG smoothing was the best,RMSEC and RMSEP was 0.5149 g/L and 0.4719 g respectively,Rcal and Rp was 0.9506 and 0.9624 resp ectively,RPD was 3.6814.The MH TA model of the unpretreated original spectrum established in the full spectral range is optimal,its RMSEC and RMSEP of 0.4720 g/L and 0.5693 g/L respectively,Rcal and Rp was 0.9119 and 0.8839,respectively.RPD was 2.1383.After paired sample T-test,the ME TA model had good predictive effect on the predictive values and the actual values in validation set,the mean values of the predicted values and the actual values measured by the traditional methods at different ripening periods,but there were obvious differences between the predictive values and the actual values in validation set,the mean values of the predicted values and the actual values measured by the traditional methods at different ripening periods in the TA model of MH grapes.3.In the visible-near infrared range of 430 to 1050 nm,the anthocyanins model of ME grapes established in the wavelength ranges of 623.4～686.2 nm,748.9～810.1 nm,and932～991.3 nm after multiple scatter correction was best.RMSEC and RMSEP was 0.0205g/L and 0.0197 g/L,Rcal and Rp was 0.8361 and 0.8791 respectively,and RPD was 2.0981.The anthocyanins model of MH grapes without pretreatment were optimal in the whole wavelength range.RMSEC and RMSEP was 0.0117 g/L and 0.0076 g/L,respectively.Rcal and Rp were 0.8545 and 0.8548 respectively.The RPD was only 1.9269.Although the anthocyanin model had certain predictive ability,the standard deviation was still large,and the predictive accuracy of anthocyanins model needed to be improved.4.In the wavelength range of 950 to 1650 nm,the model for the comprehensive scores of ME grape quality established in the wavelength range of 950 ～ 995 nm,1100 ～ 1195 nm after multiple scatter correction and detrend pretreatment was the best,RMSEC and RMSEP was0.3484 and 0.3238,Rcal and Rp was 0.8865 and 0.9023 respectively,and RPD was 2.3195.the comprehensive scores model of MH grape quality established in 950～995,1100～1145,1200～1245 nm after standard normal variable transformation and detrend pretreatment was the best,RMSEC and RMSEP was 0.3670 and 0.3930,Rcal and Rp was 0.8398 and 0.8695,and RPD was 2.3195.After the paired sample T-test,there was no significant difference between the predictive value and the true value of the comprehensive score model of ME and MH grapes,but there were still samples with large deviation.