Research For Wheat Gibberellic Disease Forecast Model

As the second largest grain crop in China,the production and consumption of wheat is huge in China.Wheat scab occurs frequently and is widely distributed,which seriously affects the yield and quality of wheat and affects the national food security and farmers'income.At this time,the prediction of wheat scab is particularly important.In this dissertation,wheat scab ear rate and meteorological factor data of each city and county in anhui province were selected as sample data to study the internal relationship between specific meteorological factors and scab,and set about establishing the prediction model of multi-factor wheat scab,which provided a basis for the estimation and control of wheat scab.The main research work and results of this dissertation are summarized as follows:1)carried out data consolidation,warehousing and data preprocessing.In this dissertation,the acquired disease head rate data and meteorological data were sorted and put into the database to facilitate the use and invocation of the experiment.Wheat scab ear rate data of several cities and counties in anhui province and meteorological data from march to may for many years were selected as the sample data,and meteorological factors were grouped according to ten days to facilitate the study of wheat scab ear rate and dozens of independent variables.2)a prediction model of wheat scab in anhui province based on multiple regression analysis was constructed.Firstly,the correlation analysis was carried out on the data of 54 meteorological factors,and the three factors with the largest correlation coefficient were screened out.Then,the correlation analysis was carried out on the cross factor terms of these three factors,and the prediction models with meteorological factors and cross factor terms were established respectively.3)a prediction model of wheat scab in anhui province based on BP neural network was established.Firstly,meteorological factors which passed the correlation test were taken as the input of the neural network,and the ear rate data of wheat scab were taken as the output of the neural network.Then,the prediction model of wheat scab by BP neural network was established by determining weight,threshold and transfer function.The accuracy of the model in the training set can reach more than 90%,and the highest accuracy in the test set can reach 68%.4)the prediction model of wheat scab in anhui province based on multi-classification algorithm of support vector machine was constructed-Firstly,the algorithm of multi-classification support vector machine should be selected to run the grouped sample data successively with the algorithm,and the accuracy of each group should be verified by using test sets.The average accuracy of the model in the training set can reach more than 90%,and the highest accuracy in the test set can reach 90.9%.5)comparative analysis between the three prediction models is carried out.In this dissertation,three methods were used to construct the prediction model of wheat scab,and the advantages and disadvantages of the three methods were compared and analyzed.Through the above experiments,this dissertation found that the meteorological data of march had little correlation with the ear rate of scab,which was in line with the actual situation that scab would not occur in March.The outbreak period of wheat scab was in the middle of April,the influence of late April was weak,and may was the full outbreak period of wheat scab.The prediction model of wheat scab was summarized and analyzed based on the comparison of the experimental results,and an ideal prediction model of wheat scab in anhui province was selected.