Research On Demand Of Emergency Supplies For Flood Under The Cooperation Of Baidu Heat Map And Multi-Layer Perceptron Neural Network

Affected by these factors of climate warming,rapid urbanization and land subsidence,the frequency of serious disasters such as storms and floods has rapidly increased.Particularly in recent years,many large and medium-sized cities in China have frequently suffered from floods.The occurrence of flood disasters gives people's daily life,mental health and even life safety a heavy blow.The primary task after the disaster is to carry out rescue duly and efficiently to alleviate the suffering of the people in the disaster area.In order to ensure that there is enough food,clothing,clean water,temporary accommodations,and timely treatments for all the people in the disaster area,the limited emergency resources should be reasonably allocated and rapidly dispatched.The basis for the allocation and dispatch is an accurate prediction of the demand for emergency supplies.However,the affected population,an important indicator,is often calculated by the census data when forecasting the demand for emergency supplies.The census data is static,and its update interval is a year,so it cannot reflect the dynamic distribution characteristics of the population in the affected area.At the same time,because of the characteristic of non-linearity and irregularity in the demand for different emergency supplies after the disaster,it often leads to the poor estimation results of emergency materials demand using the traditional statistical and analytical models.It is difficult to accurately estimate the demand for emergency supplies by using the traditional statistical and analytical models.Therefore,this paper uses the advantages of spatio-temporal data and the excellent learning ability of the neural network method to deeply research on the estimation of emergency supplies demand.The main contents are as follows:(1)Obtaining the dynamic information about the population based on Baidu heat map.This paper taken advantage of the Baidu thermal map which was space-time data in the emergency work,and the dynamic spatial and temporal distribution characteristics of the affected population were extracted,and the data including the number,the gravity and the direction distribution of the population at different time periods were obtained.(2)Estimating the emergency demand based on the multi-layer perceptron(MLP)neural network.This paper used the excellent learning ability of the method of multi-layer perceptron neural network.Based on the data obtained from multi-period flood disasters,and the indicators including the affected population,regional coefficient,seasonal coefficient,tents,quilts and folding beds,a forecasting model of emergency demand was built.(3)Dynamically estimating the emergency demand in the central city of Wuhan.This paper chosen the central city of Wuhan as the study area,and flood as the study disaster.The dynamic estimation results of the central city of Wuhan were calculated by the dynamic information about the population using Baidu heat map and the forecasting model of emergency demand using multi-layer perceptron neural network.The result showed that on July 16,2018,the population distribution in the central city of Wuhan obtained from Baidu heat map are obviously different at different time periods.The overall characteristics of the central city was that the areas of high heat and secondary heat accumulation in the morning and evening were small,and the areas in the afternoon were large,which was consistent with the rhythm of the commute.Besides,the population gravity centers in the central city of Wuhan were near the Wuhan Shahu park,and comparing with the daytime,the distribution directions and areas at night both changed significantly.Meanwhile,the forecasting model of emergency demand based on multi-layer perceptron neural network was built with the input indicators including the affected population,regional coefficient,and seasonal coefficient and the output indicators including the tents,quilts and folding beds in the emergency supplies.And it confirmed that the model had a better forecast effect by the correlation test between actual values and predicted results.In summary,it can provide a basis for the government departments to determine the amount and the location of emergency supplies,by combining the forecasting model of emergency demand with the number,the gravity center and the direction distribution of the population in the central city of Wuhan,so this paper has a certain reference value.