Quality Identification And Origin Discrimination Of Tieguanyin Tea Based On Near-infrared Spectroscopy Technology And Genetic Algorithm

Anxi tieguanyin tea was collected as the research materials in this study.Genetic algorithm(GA)was applied to special wavelength selection based on the near infrared spectral analysis technology,and then optimized and built the mathematical analysis model.The model is used to qualitative and quantitative analysis of Anxi tieguanyin tea quality,and realize the rapid and accurate origin discrimination of Anxi tieguanyin tea.The study will optimize the fast non-destructive testing system of tea quality,advance the standardized process of tea industry,as well as provide valuable reference for protecting regional special tea of China.1、Studies on the optimization of tea polyphenols rapid determination model based on Genetic AlgorithmWe built the Anxi tieguanyin tea polyphenols quantitative analysis GA-PLS model based on the genetic algorithm.Optimization results showed that,compared with full spectrum of PLS model,the optimized tea polyphenols determination of GA-PLS model became leaner and more nimble,and the data volume for building calibration model reduced from 1557 to 552 by Genetic algorithm.The prediction correlation coefficient(RP)of validation set samples increased from 0.73 to 0.97,increased by 32.9%;the root mean square error of prediction(RMSEP)reduced from 0.91 to 0.3,decreased by 67%.The quantitative analysis GA-PLS model can accurately and rapidly determinate the polyphenols of Anxi tieguanyin tea well.2、Rapid evaluation of the tea quality of Anxi tieguanyin based on genetic algorithmIn order to find a fast and non-destructive method for quality evaluation of Anxi tieguanyin tea,we built the PLS and GA-PLS calibration models based on the genetic algorithm and partial least squares(PLS).The models were used to obtain the scores of comprehensive quality and single-factor quality of Anxi tieguanyin tea.(1)The results of comprehensive quality scores determination model showed that,the PLS model displayed the highest prediction performance after the Fourier Transform near-infrared(FT-NIR)spectrum being processed by smoothing,the second derivative and normalized methods.Statistic results of PLS model:Rc=0.921,RMSEC=0.543,RP=0.913,RMSEP=0.665.NIR spectra ranged from 6670cm-1 to 4000cm-1 were selected,and the data volume for building calibration model reduced from 1557 to 408 by Genetic algorithm.Statistic results of GA-PLS model:RC=0.959,RMSEC=0.413,RP=0.940,RMSEP=0.587.The prediction precision of calibration set and validation set of GA-PLS model is better than those of PLS model.It suggested that the GA-PLS model provides strong reference and possesses promotional value.(2)This study built the single-factor quality scores determination GA-PLS model of taste,aroma and liquor color,based on the modeling methods of comprehensive quality scores determination GA-PLS model.The Rp of validation set samples of these three models were higher than 0.91.It suggested that the single-factor quality scores determination GA-PLS model has good prediction performance.Compared with the comprehensive quality scores determination GA-PLS model,the results indicated that the predicted results of the comprehensive quality scores determination GA-PLS model was the best,it can reflect the whole character of tea more comprehensive and accurate,and be applied to quality evaluation of Anxi tieguanyin tea.3、Geographical origin identification of Anxi tieguanyin based on genetic algorithm(1)In order to identify the tieguanyin tea collected from Anxi,Datian and Huaan,the genetic algorithm was applied to special wavelength selection,and combined with the support vector machine(SVM),then built the geographical origin identification GA-SVM model of tieguanyin tea with the radial basis function(RBF)as kernel function.The results showed that the GA-SVM model had the highest predictive ability when the first derivative and normalized pretreatment was employed with the penalty parameter C=104 and the nuclear parameter g=0.0075 in the geographical origin identification model.The identification accuracy of calibration set was 99.42%,validation set was 95.18%and cross validation test was 98.4%.Compared with full spectrum of SVM model,the geographical origin identification of tieguanyin GA-SVM model became leaner and more nimble.The identification accuracy of validation set samples increased by 10.84%,it can successfully identify the origin of tieguanyin tea.(2)The study built the GA-SVM model to identify the Anxi tieguanyin tea which was collected from Changkeng,Gande and Taozhou.The results showed that it had the highest predictive ability when the first derivative and normalized pretreatment was employed in the geographical origin identification GA-SVM model of Anxi tieguanyin tea.The identification accuracy of calibration set was 93.04%,but The identification accuracy of validation set was just 72%,only 36 of the 50 validation set samples were accurately judged,and 14 samples were incorrectly judged,the error was very large.Due to the slightly less experimental samples volume,the geographical distance between Changkeng,Gande and Taozhou is very close,furthermore the ecological environment and tea processing technology of these three regions is similar,accuracy of model is relatively low.As a result,further study on the methods of optimizing model need to be done,we do not recommend this model applied to the actual production process temporarily.