The Research Of Typhoon Disaster Loss Prediction

Guangdong province is in the most frequent areas of global tropical cyclone: Northwest Pacific and South China Sea area,and seriously affected by the typhoon.Disaster loss prediction is one of the important part in the decision making process of disaster prevention and reduction.But for a long time,the research of disaster loss prediction is always a weak link in the typhoon disaster prevention and mitigation in China.Disaster prediction errors caused by disaster prevention and mitigation of decision-making errors are not uncommon.Therefore,it is urgent to improve the research level of typhoon disaster loss prediction.Comprehensive,accurate,timely and scientific prediction of typhoon disaster,has important practical significance.Based on the study and research of typhoon disaster loss prediction and evaluation at home and abroad,this paper established a prediction model of typhoon disaster loss based on BP neural network,aiming at the disadvantages fliabletotrapinlocaloptimalvalueandslowrateof convergence,low accuracy of prediction,we adopt the adaptive mutation particle swarm optimization algorithm to optimize BP neural network,get the reliable prediction model of typhoon disaster losses.In this paper,the characteristics of typhoon in Guangdong Province,the selection of disaster loss assessment factors,some of the key issues in the construction of the disaster loss prediction were studied in the following aspects:(1)The selection of typhoon disaster loss prediction and evaluation factors,the socio-economic factors,the intensity of the typhoon intensity factors,including the minimum pressure,duration,the impact of the scope and so on.In particular,the wind and rain of the typhoon is described from the point of ascension to the surface,in addition to record the typhoon wind and rainfall extremes,but also the distribution of wind and rain and the scope of the situation.(2)Establish the database of typhoon disaster in Guangdong province.Database is a comprehensive database of meteorological data and disaster loss data of typhoon,as well as the social and economic data of Guangdong.From the "Guangdong Statistical Yearbook","tropical cyclone Yearbook",China Meteorological science data sharing service network,multi-channel access to raw data.(3)Research on prediction model of typhoon disaster loss.How to improve the accuracy of the model is the main research content of this paper.BP neural network and BP neural network model with adaptive mutation PSO optimization is used to predict the loss of typhoon disaster.(4)Based on the typhoon disaster loss prediction model constructed in this paper,we design and develop a typhoon disaster loss prediction system of Guangdong on Web page,in order to provide information support for Guangdong province.The prediction of typhoon disaster losses in China should be studied in the region.This paper selects Guangdong Province as the research object,through the analysis of geographical environment,climate characteristics,socioeconomic status,typhoon characteristics of Guangdong province,we select 12 evaluation factors to constructed a method for predicting typhoon disaster losses based on BP neural network and BP neural network model with adaptive mutation based on PSO optimization.The prediction results shows that the two methods can get reliable prediction results,and the BP neural network model based on adaptive mutation PSO optimization is more superior.The prediction results of the two methods can provide information support for the decision-making of Guangdong typhoon disaster reduction work.