The Research And Application Of Procurement Forecasting Model Based On Gray--Radial Basis Function Neural Network

Forecasting as a science which forecasts the future events is widely used in various areas of our lives. And the procurement forecasting which is among the forecasting can provide a scientific and rational method for us to reduce inventory, increase profits, and shorten the time better. It has a certain role in guiding. Therefore, the research of procurement forecasting has some practical significance. This paper aims at the theory of procurement forecasting to study the establishment and application of the procurement forecasting model.First, this paper briefly describes the background and significance of procurement forecasting, and the status quo as well as domestic and foreign research. On the basis, this paper describes the theory and the advantages and disadvantages of the traditional methods and modern intelligence methods of procurement forecasting. At the same time combining with the example which is selected by this paper puts forward the impacted factors of procurement forecasting.Again it mainly introduces procurement forecasting models which use two kinds of intelligent algorithms to build. Namely, the gray system theory method and RBF neural network method. To apply gray system theory method to establish the traditional GM(1,1) model and improved GM(1,1) model and using the examples to verify them. Obtain the compared results of the predicted value and the actual value, and eliciting the advantages and disadvantages of these two methods.And then combining the advantages of these two methods build gray-RBF neural network combined forecasting model. This paper's combined forecasting method is parallel connection gray-RBF neural network. Using the optimal weight combining's method. It will use improved GM(1,1) model and RBF neural network parallel combination. Obtain the compared results of the predicted value and the actual value. This combining method reduces the randomness which is implicit in the datum, and improves the prediction accuracy. Finally, use the selected example to verify the effectiveness of gray-RBF neural network method. Then compare the predicted results to gray system theory and RBF neural network predicted results. Proving this combined forecasting model is superior to a single forecasting method, and it is a feasible method.