Estimation Of Polyphenol Content In Taiping Houkui Tea Based On Hyperspectral Technique

China is the first country of producing tea in the world.The export volume of tea is large,which occupies a very important position in China's economic crops.Now,in the management of tea gardens,in order to master the growth status of tea in time,the appearance diagnosis is carried out through personal experience.Recognizing the priority of tea growth and the functional evaluation of fresh tea quality is a common method.This method is greatly affected by subjective factors,lacks reliable scientific basis,and consumes a lot of labor forces and material resources.And the result of judgment is easily affected by subjective factors and it has deviations.Therefore,it is necessary to establish a real-time,fast and high-precision quality detection method for raw tea.On the basis of comprehensively analyzing the correlation between substance content and hyperspectral characteristic band,take the content of tea polyphenols as the research object,combine the stoichiometric methods and hyperspectral techniques,use different pretreatment and modeling methods,explore the accuracy and stability of different pretreatment and modeling methods.In this study,60 tea samples were collected and divided into a validation set and a modeling set.While the tea samples of Taipinghoukui were under the real-time collection of the spectral data in the hyperspectral,picked the fresh tea professionally in the region,after the fresh-maintaining process,and then used high performance liquid chromatography to detect the content of tea polyphenols in the tea samples,homogenized the Taipinghoukui to improve the accuracy of test.In the test,the effects of the first-order derivative function,second-order derivative function,and smoothing pretreatment were compared and combined pretreatment to obtain the best pretreatment combination of the content of polyphenol in Taipinghoukui,then processed the spectral data.Experiments show that the characteristic bands under different pretreatments which were obtained by principal component analysis were 590 nm,487nm,479 nm and 536 nm of the first-order differential derivative;744nm,567 nm,691nm and 418 nm of the second-order differential derivative;632nm,401 nm,369nm,607 nm of first derivative SG;601nm,379 nm,717nm,437 nm of second derivative +SG.The correction model was established by multiple linear regression and least squares method,and the accuracy of prediction was compared and analyzed.According to the experimental data,the least squares regression model with first derivative + SG smooth fitting has the highest accuracy.The predicted correlation index of the content of tea polyphenol is 0.842,the standard deviation and root-mean-square deviation of prediction are 1.313% and 1.756%.Compared with the real value,the deviation of predicted value of tea polyphenols is still relatively small,and it can be used to evaluate the specific content of tea polyphenols in actual samples.