Comparison of prediction model for tobacco quality characteristics in central Yunnan province

Abstract:
Ecological factors are one of the basic conditions for the formation of tobacco quality characteristics. The neural network is used to predict the main ecological and chemical components of flue-cured tobacco in 2019-2020. The results showed that in the relative contribution rate of ecological factors to the chemical components of flue-cured tobacco, the average relative contribution rate of meteorological factors was significantly higher than that of soil and altitude, with an average contribution rate of 69%. The main influencing factors of different chemical components were different. The determination coefficient of multivariate statistical prediction model is about 0.30, while the neural network prediction model is about 0.80, and the RMSE and nRMSE of neural network prediction model are lower than those of multivariate statistical prediction model. Among them, the simulation effect of total sugar and reducing sugar is satisfactory, with nRMSE of 4.68% and 5.54% respectively. The study on the correlation between chemical components and ecological factors of flue-cured tobacco is of great significance to make rational use of climate resources and further improve the intelligent evaluation of flue-cured tobacco quality.