| Rice blast is an important disease on rice in Anhui Province,and the neck blast is most harmful.Accurate prediction is important for guiding the prevention and control of the neck blast.The random forest algorithm uses multiple decision-making trees to integrate self-learning,which has the advantages of fast speed,high accuracy,and strong stability.This article is based on the resistance of the varieties of rice planting areas and the current status of the occurrence and prevention of ears and neck blast.Based on the random forest algorithm,the neck blast prediction models are established to build a predictive system.Below are key research findings:1.Through Pearson correlation coefficient method,chi square test and maximum mutual information coefficient method,four meteorological factors such as relative humidity,average temperature,maximum temperature and minimum temperature are selected as prediction characteristic factors.2.Organize and analyze the disease data,meteorological data,and variety resistance data of the main rice producing areas in Anhui Province,and construct a data set of factors affecting rice neck blast in Anhui Province.Select 5 days as the meteorological data prediction time window through correlation analysis;It was found that the correlation between meteorological factors in the late stage of booting and the degree of ear neck blast was the highest.And based on the resistance distribution of rice varieties in Anhui Province,the ratio method is used to obtain resistance parameters for later adjustment of model prediction results.3.Based on the Random Forest algorithm,a prediction model for rice panicle neck blast in Anhui Province was established.The determination coefficient and root mean square error were used as indicators to evaluate the performance of the model.The determination coefficient of the established neck blast prediction model was 0.9331,and the root mean square error was 0.8141,which was superior to the support vector machine(SVM)and Light GBM algorithms.The model was validated using survey data from 2022 that were not involved in the model construction,and the prediction effect was good.4.Using a front-end and back-end separation architecture,a rice neck blast prediction system in Anhui Province is constructed,which is embedded in the Anhui Province crop disease monitoring and warning system.Users can quickly obtain the prediction results of rice neck blast by inputting meteorological data and variety resistance information in the corresponding interface,which can provide reference for the prevention and control of rice neck blast in Anhui Province. |
