Prediction Of Dry-hot Wind Disaster Of Wheat Based On Distributed Computing

Wheat is one of the main food crops in China,the healthy growth of wheat is of great significance to the steady development of our national economy.Henan province is located in the Middle East of China with winter wheat as the main food crop.Winter wheat has a long period of growth,but because of the changeable climate in north China,agrometeorological disasters occur frequently,which seriously affect the growth of wheat.dry-hot wind is the most common and dangerous,severe even can make the wheat yield reduction of more than 20%,so in the case of dry-hot wind disaster often occurs,if can be predicted before the disaster,and the warning signal released,so that farmers actively take measures to prevent disaster,reduce the impact of disasters,is of great significance to the steady growth of wheat yield.In this paper,according to the study of wheat dry-hot wind agrometeorological disaster prediction,in view of the relatively weak short-term forecast of dry-hot wind disaster in wheat and its dependence on weather forecast,this paper constructs a forecasting model of dry-hot wind disaster of wheat based on data mining technology and making full use of historical meteorological.After the model is feasible,the computational model is migrated to the distributed environment to deal with the computational efficiency and real-time problem caused by the growth of data volume.Finally,a prototype system for predicting the dry-hot wind disaster of wheat is constructed,which provides users with disaster prediction service.The work of this article mainly includes the following aspects:(1)Prediction model of dry-hot wind disaster in wheat..In this paper,the disaster mechanism of dry-hot wind of wheat is analyzed after the grade index,through the historical data collation annotation,after marking completes,the preliminary screening affects the dry-hot wind early factor,a total of 18 dimensions,uses the principal component analysis method to extract the important factor which affects the dry-hot wind occurrence,provides the solution to the dimension disaster which the model faces.Then,the BP neural network and Support vector Machine(SVM)algorithm are selected as the alternative algorithms to predict the dry-hot wind disaster of wheat,and a combined forecasting model based on BP neural network and support vector machine is constructed.Finally,the model is evaluated to prove the feasibility of the model,which can be used to predict the dry-hot wind disaster of wheat.(2)Algorithm design in distributed environment.In order to deal with the problem of real-time computing caused by the growth of data,a parallel BP neural network algorithm and a parallel support vector machine(SVM)algorithm based on distributed computing Platform(Spark)are designed.Based on the model of dry-blast-wind disaster prediction of wheat,the model and the efficiency of the model under the single environment and the distributed environment are compared.It is proved by experiment that the parallel algorithm in distributed environment is more efficient than the serial algorithm under stand-alone environment in the case of large amount of data,and the model can provide technical support for real-time and efficient computation.(3)Development of a prototype system for predicting dry-hot wind disasters in wheat.The architecture design mainly includes data acquisition and preprocessing,model design in distributed environment and real-time prediction service.The data processing mainly includes the mining of historical data and the calculation and analysis of real time data,and the real Time prediction service is based on the model design in distributed environment,the Distributed model computing subsystem is constructed,and the prediction results are displayed through WEBAPI technology.In the Distributed Model computing subsystem,the data received every day is preserved as the historical data,and the model is updated in real time to predict the next wheat dry-hot wind disaster.By collecting historical data,this paper establishes model of wheat dry-hot wind meteorological disaster,verifies its model feasibility,realizes model design under distributed environment,constructs prototype system,and effectively predicts dry-hot wind of wheat in real time.The work done in this paper provides the technical support for the users to receive the forecast information in time and take measures to avoid the reduction of production,which provides some ideas for the prediction of other meteorological disasters,and has a certain significance for the steady growth of wheat production in China.