Based On The Quantity And Safety Of Agricultural Products Of Grey Neural Network Forecast Model Research

Agricultural products quantity security is a major research field involving with the people’s livelihood and national food safety. Meanwhile, food production forecast plays an important role in food security to grasp the grain yield trend. Forecasting model optimizing and agricultural products quantity forecasting system building is proposed relied on the program "Agricultural Products Quantity Security Early Waring System(APQSEWS)" for deals with agricultural product historical data based on grey system and neural networks theory.Grey neural network model has been successfully applied in many fields with good prediction effect. Because of the strong volatility of food production statistics data, a dynamic combination forecasting model is proposed to improve the prediction accuracy based on GM(1,N) and BP neural network. In order to further promote the process of informatization of agricultural products forecasting model, the key technology of Silverlight based on rich internet applications (RIA) and Matlab is adopted to develop the integrated system of agricultural products quantity security forecasting model. The main research and achievement are as follows:With time-series and less information of agricultural products historical data, the forecasting model parameters are analyzed dynamically and the optimal initial value is selected as the basic theoretical model to establish a static timing-series prediction model GM(1,1) and BP neural network-based residuals correction. The analyzed Grey model includes GM(1,1), DGM(1,1), residuals of GM (1,1) and Verhulst model.Agricultural production is closely related with other factors in crop growth process, such as, sown area, irrigated area, damage area, fertilization area and pesticide use amount. Impact factors are analyzed to study the correlation coefficient and con-elation factor of each impact, and provide the basis for the analysis and prediction of grain yield data.According to large amount of analyses on absolute correlation degree, relative correlation degree and comprehensive correlation degree, the impact factors with heavy weight are assigned as the main infectors providing reliable basis for the data selection. A GM(l,N)and BP neural network dynamic combination forecasting model is proposed and built with the selected impact factors. This model can predict the crop yield dynamically according to the effect of the correlation factors.A MVVM expansion mode for multidimensional data exhibition based on RIA is proposed to optimize the design of multidimensional data exhibition module in data warehouse, which can solve the multidimensional pivot web system problems including poor interactivity, customization and maintain difficulty and poor performance when updating the structure.Finally, Matlab and .NET inter-operating is adopted to develop the integrated system of agricultural products quantity security forecasting model, which is flexible, efficient and user friendly.