论文部分内容阅读
本研究在收集整理全省20个县(市)测报站20多年积累的大量稻螟观察数据的基础上,通过对影响稻螟种群动态的生态因素分析,选取22个待选因子,以易于普及的PC-1500袖珍计算机为工具,应用逐步回归分析方法筛选预报因子,建立预测模式。同时辅以时间序列分析和模糊聚类分析进行补充和印证。时间序列分析采用随机过程论的处理方法,把稻螟虫情变化的时间序列分解为周期项和随机项,经叠加后用于预报;模糊聚类分析是利用样本因子数据之间的模糊等价关系进行分类预报。以上三种预报方法既可独立进行预报,又可相互补充和印证,从而提高了预报的准确度。
Based on the observation data of a large number of rice borers collected from the test stations in 20 counties in the whole province for more than 20 years, this study analyzed the ecological factors that affected the population dynamics of rice borer and selected 22 candidate factors to make them easy to popularize PC-1500 pocket computer as a tool, step by step regression analysis of screening forecasting factor, the establishment of forecasting model. At the same time supplemented by time series analysis and fuzzy clustering analysis to supplement and confirm. Time series analysis uses the stochastic process theory approach to decompose the time series of rice borer and insect pests into periodic items and random items, which are used for forecasting. Fuzzy clustering analysis utilizes the fuzzy equivalent relationship between the sample factor data Classified forecast. The above three forecasting methods can not only predict but also supplement and confirm each other independently, so as to improve the forecasting accuracy.