The Application Research On The Forecast Of Inventory Demand Based-on Back-Propagation Neural Network In The Valve Manufacturing Enterprise

With the increasing expansion of manufacturing enterprise scale and the continuing growth of consumers’personalized demand, the forecast of inventory demand is becoming more and more important and difficult. Although the management model of "zero inventory" takes some steps on some enterprises, it can’t replace traditional inventory. The forecast method of traditional inventory demand is impacted by two factors that the expanding unceasingly of enterprise scale, growing consumers personalized demand and its implementation effect and level of prediction is not satisfactory. It is one of the reasons leading to serious problems in the enterprise inventory in China. The enterprises, therefore, need to improve the forecast method of inventory demand for the adaption to development of the times, deal with the increasing expansion of enterprise scale and the continuing growth of consumers’personalized demand, which is of critical value and more efforts.This thesis takes the inventory demand of electric flangedvalve in the valve manufacturing enterprise for example based on the status quo of inventory demand forecast level in China. It analyzes deeply the characteristics of the inventory demand in manufacturing impacted by two factors. Due to the effect of two factors, the inventory demand represents the following characteristics:impact factors are more and more, the relation between impact factors and inventory demand is more complex and vague, inventory demand is more no regular nonlinearity. When deal with the question of the forecast of inventory demand, this thesis provides the BP Neural Networks to solve the new difficulties and question. This thesis analyzes the characteristics of BP Neural Networks based on its structure and theory. Then it demonstrates that BP Neural Networks is suitable to solve the question of the forecast of inventory demand from the theory and practice.Finally, the thesis applies the model to the practice, then analyzes its sensitivity and degree of accuracy deeply and in detail, proves its superiority. At the time it provides a standard process to implement Artificial Neural Networks model for the forecast of the inventory demand.