【作者】 陈伟强；

【导师】 华一新；

【作者基本信息】 解放军信息工程大学 ， 地图制图学与地理信息工程， 2009， 博士

【摘要】 烟草是叶用经济作物,烟叶质量受气候、土壤、品种、施肥和烘烤技术等诸多因素的影响。其中施肥则是决定烟叶产量、产值的第一因子,对烟叶香味的贡献仅次于品种。在一定的环境和品种条件下,施肥是烤烟优质高产的关键技术之一。目前我国的烟叶质量与美国和津巴布韦相比,还有很大差距,其中一项主要限制因子是烟区土壤的养分状况不平衡,基于GIS的变量施肥技术是解决这一问题的关键。本文基于GIS技术,开展了烟区生态环境数据库建设、烟草施肥建模、烟草施肥专家知识库构建、神经网络施肥模型构建、烟草施肥决策系统设计与开发、系统应用等方面的研究。完成的工作与创新点主要包括:1.设计了烟草施肥决策支持系统的体系结构框架。根据用户需求,确定了系统总体目标。按照智能决策支持系统的基本结构,设计了系数总体框架,以及数据库、专家知识库和模型库的框架。2.设计并建立了烟草施肥决策系统基础生态环境数据库。设计了数据库的概念模型,确定了GIS空间数据库解决方案,提出了烟草施肥数据模型,建立了数据库。3.确定目标产量模型为烟草施肥基本模型,通过南阳、平顶山和洛阳三地的烟草施肥试验,对模型中的参数进行了研究与确定,建立了烟草施肥基本模型。4.设计并建立了烟草施肥专家知识库。分别建立了产生式烟草施肥专家知识库和框架式烟草施肥专家知识库。建立了烟草施肥修正的规则和养分梯度分级施肥的知识规则。设计了基于知识的烟草施肥模型和基于案例的施肥模型。5.基于模糊数学与神经网络理论建立了烟草模糊神经网络施肥模型。6.将建立的烟草施肥模型、专家知识与GIS数据库耦合,得到烟草空间施肥模型。最后在Mapobject平台之上,建立了烟草施肥决策支持系统。更多还原

【Abstract】 Tobacco is a kind of economic crops. Its quality is influenced by many factors including climate, soil, species, fertilization and baking technologies. Among all these factors, fertilization is the fundamentally decisive one to the production and output value of tobacco. Following species, it is the second decisive factor to determine the flavor of tobacco. Therefore, under a certain environment and a given species condition, fertilization is one of the key contributing factors to high production of high-quality tobacco. At present, the quality of tobacco in China is far behind that in America and in Zimbabwe, with the imbalance of soil nutrient being one of the most limiting reasons. The fertilizing technology based on GIS is the key to solve this problem.This paper includes the investigations on the construction of ecological database in the tobacco-growing areas, tobacco fertilization modeling, the building of knowledge base of tobacco fertilization, neural network fertilization modeling and the design and development of tobacco fertilization decision-making. The achievements and new ideas are as follows:Firstly, the structural framework of decision-supporting system of tobacco fertilization has been designed. According to the users’needs, the overall objective of the system has been established. In accordance with the basic structure of intelligent decision-supporting system, the structural framework of system, database, expert knowledge base and models base have been designed.Secondly, the ecological environment database of tobacco fertilization decision-supporting system has been designed and built. The conceptual model of database has been worked out. Solution to GIS spatial databases has been determined. Tobacco fertilization data models has been put forward and database has been established.Thirdly, target output model has been set as the basic model of tobacco fertilization. Through the experiments on tobacco fertilizations in Nanyang, Pingdingshan and Luoyang, all the parameters in the model have been studied carefully. And this study has led to the establishment of the basic model of tobacco fertilization.Fourthly, expert knowledge base of tobacco fertilization, which includes production knowledge base and frame knowledge base, has been designed and built. With expert knowledge, the knowledge rules of revising tobacco fertilization and of fertilization according to nutrient level have been set up. Finally, the tobacco fertilization model based on knowledge and the model based on cases have been designed.Fifthly, on the basis of theories of fuzzy mathematics and neural network, fertilization model of tobacco fuzzy neural network has been set up. Finally, through integration of tobacco fertilization models with expert knowledge and GIS database, spatial tobacco fertilization model has been generated. Based on Mapobject platform, the decision- supporting system of tobacco fertilization has been built.更多还原