Research On Nonlinear Calculation Model Of Typhoon Disaster In South China

Tropical cyclones occur on the ocean surface of subtropical and tropical regions and are one of the most destructive weather systems.Typhoons are a kind of tropical cyclone,and it often causes most serious natural disasters before and after landing,such as storm surges,mudslides and Floods and other disasters.South China(referred to as Guangdong,Guangxi,and Hainan)is close to the South China Sea.It is one of the most severely affected areas in the typhoon.In order to construct a scientific and accurate typhoon disaster assessment and prediction model,The nonlinear computational model method is studied and constructed in this paper.The model is applied to the typhoon disaster in South China to analyze and explore the relationship between the hazard factor,the hazard affected bodies and the ability of disaster prevention and reduction and disaster.This paper has done the following research:(1)Through the study of typhoon disaster data from 1981-2016 on the coast of South China,the typhoon hazard factor and typhoon disaster were used as research index variables.The sequence important point segmentation algorithm(SIP)was used to divide 1981-2016 into 1981-1993.Four periods from 1994-2004,2005-2012,and2013-2016(referred to as periods I,II,III,and IV);Combined with two-dimensional probabilistic non-uniform information diffusion to establish the risk of typhoon disasters in South China during various periods hazard assessment model;the results show that the risk of typhoon disasters in South China gradually increased from the period of I,II,III and IV,and increased the most in the fourth period.(2)The random forest algorithm was used to select 11 disaster prediction factors to construct a typhoon disaster prediction model based on the extreme gradient grading tree(XGBoost)in the severe typhoon disasters or higher typhoon disasters of South China.The three predictions of random forest(RF),BP neural network and decision tree were compared and analyzed.The model results show that the prediction effect based on XGBoost model is better than random forest(RF),BP neural network and decision tree.It shows that the method is effective and feasible in typhoonprediction,and its prediction result can be used for disaster prevention.Provide new research ideas.(3)In order to study the law of typhoon disasters,comprehensively consider the disaster prediction of typhoon disasters or higher level typhoons disasters in South China,and use the T-S fuzzy neural network improved by fuzzy C-means clustering(FCM)to establish a typhoon disaster prediction model in South China.The Grey Relational Analysis is used to select the factors of the hazard factor and the disaster prevention and mitigation factor.The principal component analysis is used to linearly reduce the factors,and the T-S fuzzy neural network and BP neural network are compared and analyzed.The experimental results show that the improved T-S fuzzy neural network performance is better than the other two methods,with higher prediction accuracy.