论文部分内容阅读
应用相关分析方法分析了浙江省19个县1988～1999年晚稻稻瘟病发病与有关环境因子的关系,筛选了8个气象因子用于晚稻稻瘟病发生程度长期预报。根据各预报因子与稻瘟病发病程度相关性,采用邻接二维图论聚类分析法,将19个点(县)划分为4个生态区。每个生态区内运用BP神经网络技术建立模型,并进行拟合和试报。1997～1999年试报验证,在划分稻瘟病生态区的基础上,应用BP神经网络模型对稻瘟病进行长期预测预报是可行的,3年试报成功率分别是78.95％、84.21％和78.95％。文中还对该方法与过去常用的预报方法的试报结果作了比较。
Correlation analysis was used to analyze the relationship between the incidence of rice blast and the related environmental factors in 19 counties of Zhejiang Province from 1988 to 1999. Eight meteorological factors were screened for the long-term prediction of the occurrence of late rice blast. According to the correlation between each forecasting factor and the incidence of rice blast disease, 19 points (counties) were divided into 4 ecological zones by using adjacent two-dimensional graph theory clustering analysis. In each eco-zone, BP neural network was used to establish the model and fit and test the report. From 1997 to 1999, it is feasible to forecast long-term blast of rice blast by using BP neural network model based on the division of the blast area. The successful rates of three-year trial are 78.95%, 84.21% and 78.95% . The article also compares the results of this method with those of the commonly used forecast methods in the past.