Research On Grain Output Prediction Based On Grey Combination Method

Food is the basic condition for the survival of Chinese 1.4 billion population.Chinese grain output can currently achieve the goal of balance between supply and demand.However,with the complex and diversified demands of people on food,there is still a lot of pressure on Chinese food supply.At the current stage,it faces the severe test of a large population,little cultivated land,and a shrinkage of arable land year after year,which seriously affects the food security of country.Thus,it is necessary to make an effective forecast of grain output in China,study the law of changes in grain production,and make accurate predictions of its development trend.It is of great significance to ensure the national food security and assist the food administrative department in making scientific and effective decisions.In this paper,the status of grain production in China was analyzed firstly,based on uncertain characteristics of "small sample" and "poor information" of grain production data,combination model based on grey forecasting method was used to make predictions on the respective 3,5,and 8 years of food production.Firstly,a three-year short-term forecast of Chinese grain output was made by using a grey extreme learning machine combination forecasting model.The normalized processing was added to the extreme learning machine model to make the prediction more accurate.Secondly,based on the principle of residual correction,the ARIMA model was used to correct the predictions of the grey model.A grey ARIMA model was established to predict the national food production and wheat production in the next five years.The results show that the grey ARIMA combined model has better prediction accuracy than the grey model.Finally,based on the gray ARIMA combined model,a grey combined multiple regression dynamic model was proposed to forecast the grain yield for 8 years.The grey correlation method was used to analyze the main factors affecting grain yield,and the impact factors with larger correlations were selected to establish the grey ARIMA combination model,and combined dynamic combination of multiple regression model to predict the long-term food production.Through the short-term,medium-term and long-term predictions of Chinese grain output,the results show that: Gray extreme learning machine and grey ARIMA combination model enable short-term,precise food production forecasting.By comparing the single predictive model,it is proved that the combined model has better prediction accuracy and better model applicability.At the same time,the grey combination multivariate regression dynamic model also complete the medium and long-term prediction of grain yield.The algorithm has high robustness and the overall prediction effect is relatively stable.The average error is 2.95%.It has provided new technical approaches for short-term,medium-term and long-term predictions of grain production in China.