The Demand Forecast And Reserve Management Research Of Emergency Relief Supplies

China is a country which has a large population and vast territory, andwhere disastersoccur frequently. The emergency management of sudden disaster relief emergency events consists of four links, namely, four stages of prevention, preparedness, response and recovery. The Emergency Management of sudden disaster relief emergency events includeconstruction of emergency management system, construction and implementationof emergency plan, emergency supplies management, daily emergency rehearsal and so on. Among them, the management of emergency supplies in emergency management is a very important section. Since the foundation of PRC, the relevant departments of the central government and local governments according to many years of experience in disaster relief and disaster situations do some emergency supplies reserve work orderly. China's central government issued a national guida nce document of disaster relief, i.e. " overall contingency plan of national public emergency events ". But the demand of reality for emergency relief is higher and higher. There is still a big gap between our current emergency supplies reserve system and emergency relief needs. The previous studies of Chinese scholars' on emergency logistics focused on building management system, and mostly study the problem with qualitative research method. And the research for relief materials focused more on inventory control model(related replenishment issues) and the strategy selection of reserve mode, and focused less on supplier selection under the mode of physical reserve. So the most important in the article is to study the need prediction and reserve management of relief materials.Firstly, the article use RBF neural network to build earthquake prediction model of the number of casualties for predicting personnel casualties, and compare the predictive value of the RBF neural network, BP neutral network and the actual number of casualties in Yushu earthquake. Secondly, construct estimation formula based on previous research scholars' research results to predict the demand of charged emergency supplies and non-expendable class of materials for a short time after the earthquake. Again, combine the traditional ABC classification with Kraljic model, and construct the classification system ofdisaster relief emergency supplies and qualitatively select the appropriate mode according to the characteristics of each type of reserve supplies. Finally, use the operational research theory of storage to construct the inventory control model of emergency relief supplies; use the multi-objective programming method to studythe optimization decision of supplier selectionabout relief materials.The study of the article finds: on the one hand, the results of analysis prove that the model of RBF neural network has better accuracy than previous neural network method, and RBF neural network operate faster than them, and the most, RBF neural network finds the global optimal solution;the method innovation of the article is to use RBF network which has higher prediction accuracy and faster computing speed to study the problem of earthquake personnel casualties; on the other hand, using multi-objective decision method to study the supplier selection of physical reserve emergency supplies is quantitative analysis with more scientificity and operability than qualitative analysis.