Application Of Support Vector Machine In Prediction And Early Warning Of Hail Disaster

Hail disaster is a severe meteorological disasters caused by the strong convective weather systems.It appears though small,the time is relatively short,but fierce and high strength,and often accompanied by strong winds,heavy rainfall,rapid cooling and paroxysmal disaster weather process.The Xinjiang area is the frequent occurrence of hail disaster area every year,hail brings huge losses to agriculture,construction,communications,electricity,transportation and people's lives and property every year.Therefore,it is necessary for us to conduct in-depth research on hail disaster forecasting and early warning,in order to better prevent hail disaster,reduce economic losses.Akesu area is located in the middle of Xinjiang,the northern edge of Tarim Basin,its agriculture is typical irrigated oasis agriculture in arid areas,also plagued by hail disaster.Currently,the weather forecast center uses series Doppler radar echo products to monitor and forecast the strong convective weather.Strong convective weather radar echo image will show more obvious image features,according to these characteristics the strong convective weather can be forecast.This paper takes Xinjiang Akesu area Hail Radar Reflectivity images as the basic data,data mining of Hail Radar Reflectivity images,extracting discriminant index to improve the accuracy of hail forecasting.Firstly,using support vector machine(Support Vector Machine,SVM)classification principle combined with wavelet transform method,data mining of hail cloud and rain image image,obtaining a two-dimensional recognition vector as a variable training model SVM model identification,and apply this model to test the experimental sample,it obtains the discrimination accuracy;then,using SVM classification principle Combined with radar reflectivity image,puts forward two different methods of data mining,were trained in SVM model,and verify the discriminant accuracy.Finally,the above three methods are combined to get the three SVM models,verify and compare the discriminant accuracy,get the final SVM model.Among them,this paper studies the classification and recognition support vector machine,the method of hail cloud radar image data extraction,the method of wavelet transform,this paper also applied the comparison method,classification method,mathematical method,summary method and experimental method to verify the feasibility of the hail prediction.The structure of this paper: the first chapter describes the background of early warning,hail disaster prediction research significance,research status of hail disaster prediction and support vector machine at home and abroad,the research contents and methods;the second chapter discusses the hail cloud recognition model research based on SVM and wavelet transform;the third chapter discusses the research on identification of hail cloud model based on SVM and radar reflectivity;the fourth chapter,discusses the study on the method of agricultural hail suppression in Akesu area;the fifth chapter,combine with the three methods proposed by the previous section,establish hail cloud disaster forecasting warning synthesis method based on SVM;the sixth chapter,summary and outlook,summarizes all the contents of the article,and illustrates the advantages and disadvantages of the application method.