【作者】 乔金友；

【导师】 王福林；

【作者基本信息】 东北农业大学 ， 农业机械化工程， 2007， 博士

【摘要】 农业信息化建设是解决“三农”问题和新农村建设的重要手段和途径。农业机械化作为农业的主要生产力,农业机械化信息化问题也倍受关注。近几年,相关农机部门及农机经营者有条件购置和使用计算机,而且涌现了很多农业机械化信息网络以及相关技术,使农业机械化信息化水平得以提高。尽管如此,目前农业机械化生产中适用的软件相对很少,设计、开发适用软件可以有效提高农业机械化技术的应用水平及效果,进而提升农业生产的效益,提高农民的收入,同时促进农村经济的发展,同时也可以有效提高农业机械化信息化水平。本论文就是在此思想指导下,结合农业机械化生产实践要求、结合国家“863”项目及黑龙江省科技厅科技攻关项目的具体要求而设计开发“农业机械化生产专家系统(AMPES)”。设计开发AMPES不但可以满足目前农业机械化生产领域信息化发展的需求,而且可以促进农业信息化、发展数字农业、建设社会主义新农村的重要内容之一。探讨农业机械化生产专家系统设计开发的基本理论、研究农业机械化生产专家系统的功能以及设计和开发方法,又可以为农业机械化其他领域专家系统的设计与开发探明道路和方向,为进一步发展农业信息化奠定基础。本课题研究具有重要的理论和现实意义。作者查阅了大量的相关文献,掌握国内外农业机械化生产领域专家系统研究、应用动态及发展趋势,提出我国农业机械化生产专家系统存在的具体问题,并详细叙述了专家系统的概念、结构、特点以及开发方法。以此为基础,设计开发了单机版的农业机械化生产专家系统。本论文的主要研究内容包括农业机械化生产专家系统分析和总体功能设计、各模块详细设计、系统数据库设计、知识获取方式研究、知识表示方法及知识库设计、系统推理策略研究和推理机设计、系统开发与实现等几个部分。研究过程中,力求理论与实践相结合。采用自上而下的方法进行系统分析和总体功能设计、各模块详细设计;采取深入生产实际调查研究、走访领域专家、自行研究设计并摄制录像等方法,获取大量领域知识;按照层次原则设计数据库和知识库,并设计开发数据库/知识库管理系统。本论文设计和开发了人工知识获取、半自动知识获取与自动知识获取相结合的知识获取机制,采用产生式规则和语义网络方法表示本系统领域知识,从而使专家系统设计和开发过程中的“瓶颈”问题得到很好地解决。以此为基础,对推理策略进行深入分析与探讨,采用正向推理和广度优先搜索策略,引用基于实例的推理思想,设计基于产生式规则的推理机和基于语义网络的推理机;采用启发式思想,依据知识库知识组织顺序以及知识的解释说明进行冲突消解。本论文采用Visual BASIC 6.0程序开发语言和Microsoft Access数据库系统,采用原型法、自上而下、自下而上相结合的方法,开发实现了单机版的“农业机械化生产专家系统(AMPES)”。在研究过程中,取得的主要成果有:(1)通过系统分析,明确了系统的总体功能,设计了系统总体架构。进行了系统功能模块划分,并详细设计了各模块功能、工作原理及运行流程,设计了各模块数据库、知识库,为系统详细设计和开发工作奠定基础。(2)首次设计并开发了包括农业机械化作业工艺过程及规范、农业机器选型配备和农业机器技术保养三个主要功能模块的单机版“农业机械化生产专家系统”。该系统可以实现不同地区、不同种植结构和种植方式条件下农业机械化作业工艺过程制定、作业工艺规范咨询、农业机器选型与配备、农业机器技术保养规范咨询、保养计划制定以及农业机器信息浏览等功能。(3)根据系统要求,深入生产实际进行大量调查研究,获取大量的文字、图片及音像等资料,并对其进行了详细整理和归纳。同时,自行研究设计脚本、摄制并制作部分型号拖拉机技术保养操作过程的录像、拍摄大量照片,为系统设计和开发提供必备的知识,增强了AMPES知识表达直观性,提高了系统界面友好程度。(4)对AMPES知识获取问题进行了重点研究,确定的人工获取与半自动获取、自动获取相结合的知识获取机制,设计并实现了领域知识的半自动获取和自动获取功能。有效解决了专家系统设计和开发过程中知识获取这一“瓶颈”问题。提高了知识的可信度,缩短了知识获取的周期,提高了系统效率和咨询效果。(5)根据AMPES及各模块功能要求,分层次设计数据库,对综合数据库和各模块数据库的关系、层次结构、数据库中所含表的数量及表的具体结构进行了详细设计。设计开发基于自身的三层数据库的数据库/知识库管理系统(DBMS/KBMS),不但可以实现数据库/知识库的常规维护功能,而且该模块依据自身的数据库结构及内容自动加载AMPES系统所有数据库和知识库,使DBMS/KBMS程序代码完全脱离AMPES数据库而存在,提高了系统的适应性,也简化了代码维护工作。(6)根据AMPES所涉及知识的特点,采用常规的产生式规则表示农业机械化作业工艺过程、农业机器选型等领域知识;根据源作业工艺过程知识特点,探讨并采用语义网络方法表示源机械化作业工艺过程的专家知识。实际应用证明,系统研究设计的知识表示方法有效实用,便于实现系统功能。(7)研究了AMPES推理控制策略,设计了基于产生式规则的推理机,包括基本工艺过程推理机、专家选型推理机、普通用户选型推理机等;设计了基于语义网络知识的源工艺过程推理机。同时,引用基于实例的推理思想,恰当地解决了基本工艺过程推理问题,也提高了基本工艺过程知识库知识自动获取能力。(8)采用正向推理方式、广度优先搜索策略设计AMPES推理过程;采用启发式思想,依据知识库知识组织顺序以及知识的解释说明进行冲突消解。针对工艺过程制定和农业机器选型等比较复杂的推理问题,采用相关推理机串联或并联使用、逐层推理的推理策略完成推理过程设计,有效地实现了系统功能。(9)结合系统需要,设计开发了生产率法、能量法农业机器配备程序,实现了计算机自动划分作业阶段、各作业阶段负荷值的计算以及作业负荷的调整等功能,同时设计开发了基于计算法的农业机器技术保养计划制定程序。(10)本论文开发“农业机械化生产专家系统”实现了系统代码与数据库相对独立、知识库与推理机相分离。在系统数据库、知识库变化后,不影响系统功能的实现,系统适应性强。更多还原

【Abstract】 Agricultural informationization construction is a important approach to solve problems about agriculture, ruralareas and peasantry, is a important way to construct new countrysides in China. As the main agricultural productivity, agricultural mechanization information was paid much attention. In resent years, agricultural machinery departments and operators have abilities to buy and use computers. Many information websites and correlative technology have been putting into use. All of these can enhance the level of agricultural mechanization information. Even though, the amount of the effective softwares for agricultural mechanization cannot satisfy the demand. Under these circumstances, to design and develop the effective software can promote the applying level and effect of agricultural mechanization technology, can promote the agricultural benefits, rise the peasantry’s revenue, meanwhile, can promote agricultural economic development. The agricultural mechanization production expert system (AMPES) was designed and developed under the guidance of this thought and according to requests of the national "863" research plan and key projects of science and technology office of Heilongjiang province.To design and develop AMPES can satisfy the demand of the field of agricultural mechanizing production, but is also one of the important content of promoting agricultural information, developing digital agriculture and constructing new countryside. To discuss the basic theory, function and methods of designing and developing AMPES can lead the direction for designing and developing other expert system in the field of agricultural mechanizing production, can make a foundation for developing agricultural information. This research has important significance both in theory and practice.Researching and applying status and developing trends of expert systems in internal and external agricultural mechanization production field were studied on the basis of a mass of related literatures. Problems of agricultural mechanization expert systems applied in our country were pointed out. Concepts, the structure, characteristics and developing methods of the expert system were also detailedly discussed in the dissertation. Main contents of this research include general function design of AMPES, detailed design of modules, design of system databases and research of knowledge acquisition strategies, research of knowledge presentation and knowledge bases, design of management systems of databases and knowledge bases(DBMS/KBMS), research of inference strategies and design of inference engines, development of AMPES and so on.Combining theories with practice is one of the main principles of this research. System analyzing, general designing and modules detailedly designing were accomplished using the top to down method. Great deals of field knowledge had obtained by methods, such as in-depth investigations and studies, visiting field experts, designing and producing videos, and so on. Databases, knowledge bases of AMPES were designed according to hierarchy principle, and DBMS/KBMS were also designed and developed under the same principle. Knowledge acquisition (KA) mechanism had designed and developed which combined artificial KA, semiautomatic KA and automatic KA, and rule-based and semantic network methods were used to represent field knowledge after relative research. These research work effectively solve the bottleneck problem in expect system designing and developing course. On basis that, inference strategies had been analyzed and discussed in depth, forward reasoning and width-first search strategies were adopted, the case-base reasoning thought was also introduced to design inference engines. Conflicts could be resolved according to organizing order and explanation of knowledge with heuristic thought. Visual Basic 6.0 and Microsoft Access were adopted to develop Windows-based AMPES with the methods of prototyping, top-down and down to top.Main conclusions of the dissertation are as follows:(1) The general structure of AMPES was designed and the modules were compartmentalized on the basis of the analysis of the system function. Functions, operating thought and process of each modules were designed detailedly. Databases and knowledge bases were designed and their relation to system function is explicit. All of these are the foundation for the further development of AMPES.(2) AMPES was designed and realized for the first time, which includes three main functional modules which are mechanized technology process and operating criterions module, agricultural machinery selecting and equipping module and agricultural machinery maintenance module. AMPES can realize all functions under different area, different plant structure and sow manner.(3) Knowledge resources of AMPES were studied. Mass of character, photo and video materials were obtained after consulting and dealing mass of literatures, making many investigations in the practice. Field knowledge of AMPES is abundant and represented in variety type. videos about the some types of tractor technology maintenance were produced on the basis of scripts designed by the item group according to the system requirement and lots of photos were shoot. These videos and photos can make knowledge of AMPES more abundant, can enhance the knowledge presentation more intuitionistic and can make interfaces more friendly.(4) Knowledge acquisition (KA) of AMPES is one of the important researches of the dissertation. Semiautomatic KA and automatic KA mechanism were designed. The semiautomatic KA can be realized after knowledge engineers manage knowledge bases according to suggestions provided when experts operated AMPES. Knowledge can be added and renewed automatically after experts modify the consulted results directly and then saved in knowledge bases. Before using the system in each area, running the module can effectively avoid shortcomings of the inaccurate, incomplete knowledge and other disadvantages in the course of investigation. The credibility of the knowledge can be improved, the cycle of knowledge acquisition can be shortened and system efficiency and advisory effect can be improved under the KA strategies designed in the dissertation.(5) Database design and DBMS/KBMS are characteristics of the system. According to AMPES and the function requirements of each module, database were design in hierarchical, and relations between integrated databases and module databases, hierarchical structure, the quantity of tables in database and the specific structure of tables were designed detailedly. and DBMS/KBMS was designed and developed based on the three-layer databases, not only can make the conventional maintenance functions of database / knowledge base come true, but the module can automatically load all databases and knowledge bases of AMPES through the structure and content of its own databases. DBMS / KBMS code existing independently, can enhances the adaptability of the system, and simplifies code maintenance work.(6) According to the characteristics of field knowledge AMPES involves, rule-based method was adopted to express field knowledge such as technology process of mechanized operating and agricultural machinery selection. According to the characteristics of source technology process, the semantic network method was discussed and adopted to express the expert knowledge of the source technology process. Practical application of the system proved that knowledge representation methods designed are effective and practical, are propitious to realize system function.(7) Inference strategies of AMPES were studied. Rule-based inference engines were designed such as basic technology process inference engine, expert’s agricultural machinery selection inference engine, ordinary user’s agricultural machinery selection inference engine. Source technology process inference engine was designed, which is based on semantic network knowledge representing method. Meanwhile, case-base reasoning thought was also introduced to solve basic technology process inference problem appropriately, to improve the ability of automatic knowledge acquisition.(8) AMPES inference process was designed adopting forward reference and width-first search strategies. Conflicts can be resolved adopting the heuristic thought according to the knowledge organization order and the knowledge explanations. Some complicated inference problems, such as the technology process making and machinery selection, inference processes were realized by means of the strategies using the different inference engines in series in parallel. AMPES functions were realized effectively adopting these strategies.(9) According to the system demand, programs to calculate the amount of agricultural machinery were designed and realized. Functions, such as operating period compartmentalizing automatically, operating loads calculating in each operating period and operating load adjusting were also realized. At the same time, program for making farm machinery maintenance plan was also design and realized based on the calculation method.(10) AMPES can well adapt to changes of databases or knowledge bases because it was realized under the principle of independently programming. AMPES has more advantages according to the production tests in practice, such as perfect functions, friendly interfaces, rapid knowledge acquisition, convenient operation and maintenance, applicable knowledge representation strategies and effective inference engines designed reasonably and so on.更多还原