【作者】 张沙沙；

【导师】 丁克坚； 朱诚；

【作者基本信息】 安徽农业大学 ， 植物病理学， 2013， 硕士

【摘要】 本试验应用CBR方法建立了以时间序列为轴线的病虫滚动预测模型,利用安徽省农作物病虫数据库对稻、麦主要病虫害的影响因子筛选、参数优化,确定稻、麦各主要病虫CBR预测模型因子、参数构架；并研究构建了安徽省小麦、水稻病虫害识别诊断知识库。农作物病虫害监测预报是进行病虫害综合防治的基础。本文提出了一种用于稻麦病虫害预测的新方法——基于案例推理的时间序列数据的相似年分析(CBR),并主要对其预测因子、参数进行了优化筛选。基于案例推理的方法就是通过搜索曾经成功解决过的类似问题,比较新、旧问题之间的特征、发生背景等差异,重新使用或参考以前的知识和信息,达到最终解决问题的方法。结合安徽省各生态区域的气候特点、地理位置及各病虫害的发生情况,用CBR对稻麦主要病虫害的预测预报研究,首先利用各代表县的年报历史资料建立时间序列历史案例库,再按照指定或预先设定因子匹配规格和权重参数来构建相似度函数,然后逐一计算目标案例与历史案例的匹配程度,最终选取与目标案例相似程度最高的历史案例的结论作为预测结论；在预测过程中,未来相关的预报数据也可加入目标案例的时间序列中参与计算。本研究分别筛选出了小麦赤霉病、小麦白粉病、小麦纹枯病、小麦蚜虫、水稻纹枯病、稻瘟病等主要病虫害的CBR预测因子及参数,经回检,小麦赤霉病、小麦纹枯病、小麦蚜虫、水稻纹枯病、稻瘟病的预测准确率皆达到80%以上,小麦白粉病的预测准确率为77.05%。其中,小麦赤霉病大流行年份的预测准确率达70%左右。因此,该预测参数具有一定的可靠性,可作为CBR模型预测稻麦病虫害的依据。利用Visual Basic6.0研究构建了安徽省稻麦病虫知识库,为构建该知识库,收集了安徽省稻麦常见病虫害的文字描述资料和图片资料,包括不同病害在农作物不同部位的发病症状描述、典型症状图片及常规的防治方法等,通过计算机语言将文字和图片有机地结合起来,使得稻麦病虫害的识别诊断更加直观、简易。用户登陆系统后进入主菜单,根据自己的需要,从菜单中选择所要执行的项目或输入必要的信息。该知识库具有图文并茂、种类较全、通俗易懂、操作简单、实用性强等特点,为基层植保工作人员的田间识别诊断提供了可靠依据。更多还原

【Abstract】 A rolling forecasting model of crop diseases and pests was studied using CBR theory with the time series and a system for diagnosis of wheat and rice diseases and insect pests of Anhui province was developed in this paper. Impact factors and parameters of plant diseases and pests in Anhui province were optimized.Monitoring and prediction of crop pests and diseases is the basis of integrated pest management. A New method, based on case-based reasoning (CBR) in time series data similarity analysis, was studied used in the forecast of crop diseases and pests. The case-based reasoning method solves a new problem by searching the previous similar problems and comparing characteristics and the background between new and old problems to get knowledge and information.The CBR forecasting of main diseases and insect pests of rice and wheat was studied combining with weather characteristics of various ecological regions, geographical location and the happening situation of pest. First, established a time series data case, then constructed the similarity matching specifications and weight parameters according to the specified or preset factors function, and calculated the agreement rate of each historical case, finally selected the history case with the highest agreement rate as prediction conclusions. In addition, the future forecast data can also be added to the target case. CBR forecasting factors and parameters of Wheat scab, wheat powdery mildew, wheat sheath blight, wheat aphids, rice sheath blight, rice blast were optimized, with the percentage of the results’ corresponding to historical data were above80%. And the agreement rate of Wheat scab of great outbreak was about70%.The predicted parameters have certain reliability and can be used as the basis for CBR models to predict crop diseases and insect pests.Knowledge base of diseases and pests of rice and wheat in Anhui province was set up by means of Visual Basic6.0. Written materials and picture data of common diseases and insect pests of rice and wheat in Anhui province were collected, including symptom pictures of different diseases in different crops. Combine the written materials and symptom pictures by computer language to make the recognition and diagnosis easier. Running the system in properly condition, choosing the needed item or inputting necessary information, the users could achieve satisfactory running effect. The knowledge base can offer a reliable basis for the plant protection staff with its characteristics, such as complete types, easy to understand, simple operation, strong practicability.更多还原