Study On Rapid Evaluation For Grades Of Dianhong Tea

For the purpose of an objective and rapid evaluation of Dianhong tea(Yunnan black tea)grades,different grades of Dianhong tea samples were collected to acquire near-infrared spectroscopy and hyperspectral images.Multivariate data analysis method was used to establish Dianhong tea discriminant model.Moreover,an attempt was made to explore new ways to build olfactory visualization system based on gas sensor array.The main results were as the following:(1)Partial least square discrimination analysis(PLS-DA)and support vector machine(SVM)discriminant models of Dianhong tea grades were established by comparing the images collected by Fourier transform near infrared spectrometer(FT-NIRS)and Smart near infrared spectrometer(Smart-NIRS).The experimental results showed that the modeling effect of SVM is better than PLS-DA model under the same conditions,especially for the Smart-NIRS.Taking comprehensively into account factors of convenience and cost,the discriminant model was established based on Smart-NIRS for screening different variables.Among them,the prediction set discrimination rate of GA-PLS-DA model was the highest,reaching 90.00%,which could realize the classification of Dianhong tea samples.(2)The discrimination grading model of Dianhong tea was established by spectrum and image information.Based on the principal component analysis of hyperspectral images,the four dominant wavelengths were selected as spectral features by the loadings of the first three principal component images under the all wavelengths.The gray statistical moment was used to extract 26 texture characteristic values.The correct discrimination rate of partial least squares model was increased to 94.38%,and the model established by SVM is the highest,reaching 100%.And the results showed that data fusion has been able to discriminate grades of Dianhong tea.(3)An olfactory visualization system based on gas sensor array was constructed.The sensor array was constructed by combining 27 gas-sensitive materials(including 10 porphyrins,metalloporphyrins,and 17 pH indicators)with different based materials.By exploring the differences between different based materials and spotting methods,difference images before and after sample reactions were obtained.RGB characteristic variables were extracted from difference images for integrated analysis.Filter paper as the base materials and multi-channel adjustable range pipette as the spotting method were selected due to their stability and feasibility.This study introduced a simple way of gas sensor array production,thus provided a feasible way of tea grades evaluation by olfactory color card.