Hyperspectral Estimation Model of Pu’er Tea Leaves Biochemical Parameters in Xishuangbanna

Taking Xishuangbanna Pu’er tea as the research object, ASD Field Spec 3 was used to collect the hyperspectral data of Pu’er tea leaves. The spectral data was processed by derivative spectral analysis technology and determine the corresponding nitrogen and theanine content in the laboratory. The correlation between biochemical parameters of Pu’er tea leaves and the original spectrum, 1^st derivative of spectrum, 1^st derivative of the spectrum logarithm and hyperspectral feature variables were analyzed, and a Back Propagation Neural Network model for content of nitrogen and theanine estimation was established. The results show that the correlation between the biochemical parameter content of Pu’er tea leaves and the original reflectance of hyperspectral is weak, but the correlation with the 1^st derivative of spectrum, the 1^st derivative of the spectrum logarithm and the hyperspectral feature variables are strong. The estimation accuracy of BPNN model optimized by genetic algorithm is better than that of the common BPNN model for estimating the biochemical parameters of Pu’er tea leaves. The BP neural network model optimized by genetic algorithm is superior to the common BP neural network model in estimating the biochemical parameters of Pu’er tea leaves. The accuracy of the determination of theanine content is 0.21 mg/g, R² is 0.73, the estimation accuracy of nitrogen content is RMSE of 0.36 g/kg, and R² is equal to 0.88.