Prediction Models of Nicotine Content in Flue-cured Tobacco in"Qingjiangyuan"Area Based on Meteorological Elements
Liu Jun
Chen Zhenghong
Liu Yang
Li Jingcheng
Deng Pingrang
Abstract：In order to establish highly accurate and stable prediction models for nicotine content in flue-cured tobacco in the"Qingjiangyuan"tobacco areas,this article analyzes the relationship between the nicotine content in the 2007-2021"Qingjiangyuan"tobacco and the meteorological elements during the field period,and develops two mechanism algorithm models based on linear equations and three intelligent algorithm models based on BP neural networks.Then,the performances of the models are evaluated by the Taylor plot and the rand function.The results show that the diurnal temperature range during the mature stage is the most influential meteorological factor affecting nicotine content in the"Qingjiangyuan"tobacco.Specifically,a larger diurnal temperature range in the mature stage can lead to the increased carbohydrate accumulation,which in turn inhibits the synthesis of nicotine.The three intelligent algorithm models have superior predictive performance compared to the two mechanism algorithm models.The correlation coeffi-cients between the predicted values of the BP neural network model and GA-BP neural network model and the actual values are 0.920 and 0.932,respectively,and the root mean square errors are 0.134 and 0.113,respectively.These findings indicate that these models can be used to predict the nicotine content in the"Qingjiangyuan"tobacco.Moreover,the GA-BP neural network model optimizes the parameters of the BP neural network by the GA algorithm,which improves the prediction effect and stability to a certain extent.
Keywords：Qingjiangyuannicotine contentprediction modelGA-BP neural networkassessment
Publication Date：2023-12-28
Online Publishing Date：2026-09-12(First online date of this platform, not the publication date of the document)
Pages：7( 32-38 )