Green Tea Adulteration Detection And Android Application Development Based On Micro NIRS

In recent years,the management system of China's tea market has been gradually improved,but it is still dominated by the traditional wholesale business model,which makes it difficult to maintain the market order.Some unscrupulous manufacturers add sugar or glutinous rice flour to green tea processing to facilitate its shape and flavor.These adulterates not only increase the risk of mildew,but also unfriendly to diabetics.At present,tea quality analysis mainly depends on chemical analysis and sensory evaluation.Chemical analysis methods are complicated and require professional inspectors.While,sensory evaluation is subjective and difficult to accurately detect adulterates in tea.The near infrared spectroscopy(NIRS)provides a potential solution for the identification of adulteration of tea leaves.However,most NIRS-based tea quality studies adopt desktop NIR spectrometer,which has some shortcomings such as poor mobility,high cost and unsuitable for online analysis.Based on the above,this paper carried out the green tea adulteration detection and application(APP)development based on micro near infrared spectroscopy(Micro NIRS).The main research contents and conclusions are as follows:1.Study on adulteration detection of green tea based on micro near infrared spectroscopyIn this study,a miniature near-infrared spectrometer was used to scan the samples and obtain the spectral information of all the samples,and a multi-layer algorithm model was designed to process the spectral information.This algorithm model includes three stages:preprocessing,qualitative and quantitative.In the pretreatment stage,standard normal variable was used to preprocess the spectrum by comparison.In the qualitative stage,principal component analysis is used to determine the similarity between sample types and to illustrate the classification trend.Then,partial least squares discriminant analysis(PLS-DA)and support vector machine(SVM)were used to establish a qualitative model to distinguish pure tea from adulterated tea.The results show that the SVM algorithm is more accurate than the PLS-DA algorithm for the classification of the three groups of samples,and the classification accuracy of the SVM model for the prediction set is as high as97.47%.In the qualitative stage,successive projections algorithm,and iteratively retaining informative variables(IRIV)were first used to screen the characteristic wavelength.Then partial least squares regression and support vector machine regression(SVR)were used to establish a quantitative model to predict the concentration of dopants.The results show that the SVR model simplified by IRIV algorithm can not only ensure the identification accuracy,but also effectively reduce the number of variables.The results show that the combination of Micro NIRS and stoichiometry can effectively analyze the sugar and glutinous rice flour in green tea qualitatively and quantitatively,so it is feasible to use this technology in the detection of green tea admixture.2.Development of green tea adulteration detection APP based on miniature near-infrared spectroscopy equipmentIn this study,the reflective Micro NIR spectrometer of Texas Instruments was finally selected,and based on this equipment,the analysis APP of tea sugar content and tea glutinous rice flour content analysis APP were designed and developed.After screening and optimizing the data processing method,the quantitative analysis model of genetic algorithm-partial least squares doped with tea samples was finally selected to improve the rate and stability of APP detection.With the help of Bluetooth wireless communication module,the spectrometer can be controlled by a smart phone.Users can see the spectral information such as intensity,reflectivity and absorbance through the phone interface,and calculate the spectral information by writing the prediction model of APP to realize the real-time display of doping content.The results show that the APP has good detection accuracy and stability for both modeled and unmodeled samples,and the error is less than3%.The results show that the APP has good detection accuracy and stability for both modeled and unmodeled samples,and the error is less than 3%.The research shows that the device and APP can be used to quickly detect the doping content of green tea.