A Research Of Dynamic Demand Forecasting Model For Large Earthquake Emergency Supplies

China always has lots of serious earthquake disasters. In recent years the largeearthquake happened a few times with great casualties and property losses makespeople pay more and more attention to the research of earthquake relief. Amongthem the ability of achieving disaster information is an important measure. Thisarticle aims at those large earthquake disasters and chooses those relief supplieswhich have a close relationship with the population involved, expecting to predictthe dynamic demand of emergency supplies. Using an indirect prediction methoddesign a set of prediction schemes. First, the article establishes a similarearthquake case set. Then it uses a recursive neural network based on time series topredict the number of the earthquake daily death. Finally, it calculates the dailymaterial request of the disaster areas according to the population and emergencysupplies demand formula.This paper first chooses the seismic attribute set, and then judges the similarcase sets as the sample data through calculating the Euclidean distance between thecases and the goal. The researcher then compares a variety of time seriesforecasting method, choosing the recursive neural network as it can make dynamicprediction. The paper combines the time series prediction and recursive neuralnetwork forming the recursive neural network based on time series predictionmethod. And discuss the neural network generalization ability, make its applicableto new sample.In order to ensure the supply of emergency demands, puts the safetystock also into consideration, and imports the lead time into the material needsformula.Finally, use the WenChuan earthquake happended in2008as an example, thispaper checks the validation of the model, and gives the prediction results andcompared with the actual data, at last realize the dynamic demand forecast ofemergency supplies, provide a guide to the relief work.