Research And Application Of Grain Yield Prediction Model

Grain is crucial to the growth of the Chinese economy,social stability,and the implementation of the strategy for sustainable development.Currently,COVID-19 is still active,extreme weather or natural disasters such as drought and heavy rain continue to occur frequently in countries worldwide,the threat to global food production is growing,and international food prices remain high.Therefore,China’s food security is dependent on domestic economic and social development and stability and on international food supply,demand,and prices.Therefore,grain is vital,and grain yield is necessary for adequate food security management.Prediction technology for scientific forecast and evaluation for grain production can effectively prevent and improve unreasonable problems in food production.Therefore,it has practical significance for the scientific deployment of agricultural production and agricultural policy formulation and adjustment and for the nation’s efforts to promote grain production,improve food production,and ensure food security and stability.This thesis developed a single-factor prediction model and a multi-factor prediction model.The prediction interval was separated into short-term,medium-term and long-term predictions.The research was conducted on a grain yield prediction model,providing a basis for individuals to select an appropriate prediction model based on their individual needs.First,modeling work was conducted using kernel regression,GM(1,1)grey prediction,BP neural network and partial least squares as four prediction methods.Then,China’s total grain yield was predicted,and the prediction errors of various models’ total grain yield prediction methods were compared.In addition,these four methods were used to predict China’s rice yield.Overall,this thesis investigated the effect of rice yield prediction using four distinct methods.Ultimately,the optimal grain yield prediction model was selected after a comparative analysis.The empirical analysis demonstrated that the optimal short-term and medium-to-long-term prediction models for China’s total grain yield and rice yield were consistent.The GM(1,1)grey prediction model was effective for short-term grain yield forecasting,while the partial least squares model was effective for medium-to long-term grain yield forecasting.Not only did the GM(1,1)grey prediction model have a high level of fitting accuracy,but it could also describe the future development trend of grain yield and had a high level of prediction performance.In conclusion,the optimal GM(1,1)grey prediction model was used to forecast China’s grain yield over the next three years,providing government departments with a practical reference for enhancing grain macro-control and formulating scientific planning.