| Dry-hot wind is one of the main agricultural meteorological disasters that seriously affect grain filling of winter wheat in northern China.The weather disasters of wheat dry-hot wind will bring great harm to wheat growth in the early twenty to thirty days of wheat filling.When serious dry-hot wind occurs,it can cause 10%-20% decrease in wheat yield.Henan Province is a major wheat production base.According to the Henan Survey Team of the National Bureau of Statistics,the wheat planting area and total output in 2022 ranked first in China.Hua County wheat grain output in Henan province ranks the first,was rated as the grain production advanced county,"the granary of northern Henan" said.Therefore,it is very necessary to accurately predict the occurrence of dry-hot wind of winter wheat in Hua County,Henan Province,and take effective preventive measures in time to reduce the impact of dryhot wind on wheat yield,which is necessary to stabilize the granary in northern Henan,China,and also to provide technical support for disaster prevention and reduction of agricultural production.The main contents of this study are as follows:1.At present,there are few researches on the annual type of dry-hot wind in wheat and few forecasting methods.Based on this,in order to better early warning and prevention of dry-hot wind,it is a key point to accurately predict the year type of dry-hot wind in wheat.In this study,the meteorological data of Hua County from May 13 to June 10,2000 to 2022 were used to build a Markov prediction model based on the characteristics of "no after-effect" of Markov principle.Then,the prediction results of dry-hot wind in Hua County during 2003-2022 were tested by back generation test,and the same data was selected to compare the results with BP neural network model.The results show that the prediction probability of Markov model is 75.00% with high accuracy.Moreover,based on the same data,the performance of the prediction model is better than that of the BP neural network model.Therefore,the Markov model can better warn the wheat dry-hot wind and play a role in disaster prevention and resistance,which is of great significance for improving the wheat yield.2.In order to predict the dry-hot wind disaster of winter wheat in Hua County,Henan Province,the training set and the test set were divided according to 8:2 according to the meteorological disaster index of dry-hot wind and the meteorological data of more than 40 years from 1981 to 2022.The correlation coefficient in filtering feature selection and the variable importance of random forest algorithm in integrated feature selection were used for meteorological feature selection.The algorithm improvement,parameter optimization,prediction model construction and model evaluation were carried out from the perspective of weighted random forest.The results show that the determination coefficient R2 of the test set of the improved model can reach 0.9926,which is higher than that of the traditional random forest model by 0.0229,RMSE and MAE decrease by 0.7496 and 0.6104,respectively.The coefficient of determination R2 is 0.0354 and 0.0446 higher than that of SVM and KNN models with better performance in current studies.3.The dry-hot wind disaster prediction system of wheat was established.The data needed for model prediction was obtained through early data collection and later data crawling.The evaluation and analysis of the year type of wheat dry-hot wind were realized in combination with the classification standard of the year type of wheat dry-hot wind.The WEB framework uses the popular VUE framework for development,the page design uses Element UI framework,and the page interaction uses JS.The data layer uses My SQL cluster to store and manage the background database of data,and the service layer uses Flask framework.The system visualization is mainly based on the Echarts framework of Mars3 D three-dimensional scene visualization platform for chart display.Users can view the results in real time using the operating system function. |
PDF Full Text Request