Demand Forecasting Of Large-scale Earthquake Emergency Materials Based On PSO-BP Neural Network

Since ancient times,frequent earthquakes in China have caused serious damage to China's economic level and people's lives.After a large-scale earthquake occurs,serious casualties will be caused,and a large amount of emergency supplies will be needed in the disaster area.However,because the earthquake usually occurs quickly,emergency department cannot know the casualties in the disaster area in the first time,and therefore cannot determine the needs of emergency supplies,which will make the distribution of emergency supplies to a large extent blind.Too much supply will cause serious waste of materials,while too little supply will not meet the needs of people in the disaster area.Therefore,it is very important to be able to accurately forecast the demand for emergency materials in time after the earthquake,which can not only ensure the smooth progress of subsequent rescue work,but also provide convenience for emergency departments to make decisions.This paper mainly builds a forecasting model for large-scale earthquake emergency supplies demand based on particle swarm optimization and BP neural network.The BP neural network improved by particle swarm optimization is used to forecast the death rate and injury rate of large-scale earthquakes with magnitude 6 and above,and obtain the number of casualties and the number of survivors.Then quantitative relationships between the number of survivors,the number of injured and different materials are used to estimate the demand for different materials.In this paper,the data of 20 large-scale earthquakes in the past years are selected as samples,and the training samples,test samples and variable samples are randomly assigned.When forecasting the mortality and injury rate,this paper selects 8 prediction indicators,and uses all samples for principal component analysis to finally select 6 indicators for forecasting.Then the network structure and a series of parameters are set,and the BP neural network improved by particle swarm optimization algorithm is constructed,and the optimized initial weights and thresholds are obtained using training samples.Afterwards,the network is trained and tested with the three assigned samples,which verify the validity and accuracy of the model.When estimating emergency supplies,this article mainly estimates the demand for food,cold items and medical items.By forecasting the demand for large-scale earthquake emergency materials,it can provide a certain reference for the government emergency department to determine the supply of materials,so that it can avoid the situation of oversupply or in short supply to a large extent,and meet the materials demand in the disaster area as much as possible,it can also save materials,which is conducive to the smooth progress of subsequent rescue work.