【作者】 姜海燕；

【导师】 曹卫星；

【作者基本信息】 南京农业大学 ， 生态农业科学技术， 2007， 博士

【摘要】 作物系统模拟技术是农业信息学研究的关键技术之一。作物模型研究的深入和拓展对作物模型构建方式、效率、质量和工具等提出了新的要求。借助现代人工智能和软件开发技术,探索作物模型组织、集成、共享和管理的新模式,有助于研制面向农学家的智能化建模工具。促进作物系统模拟的创新发展。本研究的目的是融合多学科的理论与技术,以作物生长系统为对象,基于作物模型资源与集成构造思想,解析与整合农作物生长模型的建模流程、模型结构、基本算法、参数类型和模拟数据,提出模型资源组织、储存、共享、集成与管理的关键技术,构建作物模型资源仓库(CMRW)及基于资源融合的作物集成建模系统(ICMCS),从而为实现作物系统模拟的智能化和自动化奠定基础。将基于资源的集成制造思想引入作物模型领域,构建了作物模型资源(CMR)、作物模型资源仓库(CMRW)和作物模型资源集成构造(ICMRC)的概念、特点与内涵.阐明作物模型资源由作物建模知识资源和软件开发资源组成,具有可管理、可复制、可重用和可迁移的特点。作物系统模拟是作物模型资源在不同状态空间的动态转换过程,基于资源融合的集成构造为作物模型资源化提供技术支撑,可以有效促进建模知识资源与模型软件开发资源的自动获取、动态存储与转化。指出研制基于资源融合的、具有学习能力、自适应和组件化特征的作物集成建模系统,有助于研究人员在非编程环境中构建与积累具有统一组织的、易于共享复用的作物模型资源,降低模型构建与开发难度,提高模型开发效率和质量。将本体方法引入作物系统模拟领域,以作物建模为主线,综合分析小麦、水稻和棉花等作物的生长模型及模拟系统,分层次建立了仿真、作物模拟和作物模型等本体,进而构建了由作物建模和作物模型内外知识框架组成的作物系统模拟框架.明确作物模拟模型和仿真本体是作物模拟本体的基础,以作物系统模拟活动本体为主线,作物、资料、模拟方法和模拟任务等本体相互协作,可形成作物建模外部知识框架.基于作物生长规則和模拟模型组分,结合作物模型本体可构建作物生育期、生物量、器官建成、分配与产量、水分及养分平衡等模型或子模型內部知识框架.作物系统模拟框架具有完整的知识表达力,提升了作物模型的共享与重用价值,为设计可重用的集成建模软件体系结构奠定了基础。以作物系统模拟框架为基础,结合模型表示与组件方法,提出了作物模型资源映射机制,构建了作物生长元模型和元数据,确立了作物模型集成构造原理和集成技术。表明作物模型可映射为作物模型概念与或树、模型算法与或图以及实现形态的综合体,模型资源可表示为“骨架+规则+算法+模板+组件”新型实体.基于作物模型及子模型內部知识框架、软件实现形态和数据的综合,构建作物生长元模型和元数据。基于组件框架、模型描述脚本、反射式组件组装、动态知识集成和模型解析等技术,作物模型资源集成机制可实现模型结构、模型组件、算法知识和模型参数的综合集成.将作物模型构建活动转化为由作物生长元模型和元数据、建模学习机制、模拟支撑工具、模型定制机构和模型构造引擎共同协作完成的学习和实践活动。作物模型资源映射与集成构造技术研究,实现了作物建模知识级与构件框架级的复用。以关系型数据库技术为基础,从概念关系出发,统一组织与融合作物模型构件库、建模知识库和实验数据库信息,采用虚节点方法扩展作物模型结构,利用“知识页+算法知识组+算法单元”方式组织建模知识,结合刻面分类方法描述模型组件属性,将作物模型映射为E-R模型,并综合基础数据库和方案数据库信息,构建了由作物模型框架库、算法库、实验数据库和方案库等共同组成的CMRW关系数据模型。并以模型方案库为线索,采用“三层资源获取”策略和算法,实现对作物模型框架、算法和参数信息的获取.基于综合数据库平台,建立了小麦生长模型资源仓库,验证了构建与存取技术的可行性,为作物模型集成与重用积累了基础资源。以作物模型资源集成构造技术为支撑,作物模型资源仓库为基础,基于.Net平台,使用C#语言和Mierosoft SQL Server数据库,研制了基于资源融合的作物集成建模系统(ICMCS),并实例应用于小麦和大麦的建模实践。ICMCS拥有层次化、组件化和智能化特征,由生育期、生物量、器官建成、分配与产量、水分及养分平衡等子模型框架、模型和模板等组件组成。基于作物生长逻辑模型和建糢服务机构,实现了模型构建、模拟运行、模型保存与输出、模型分析与评估、模型资源检索、模型字典与工具等功能。应用测试显示,ICMCS具有较强的多功能性、自适应性和可扩展性,可替换基于代码编程的传统作物模型构建方式,农学家可通过使用学习、定制、设计、配置、组装、评估和扩展等手段,实现无编程、智能化地构建、开发、积累与测试作物模型资源和初步模型。更多还原

【Abstract】 The crop system simulation is one of the key technologies in agricultural information studies. With the expansion of crop model research, new requirements about the method, efficiency, quality and tools for crop model construction were proposed. For developing agronomist-oriented intelligent modeling tools and promoting the innovation and development of crop system simulation, it is useful to adopt the technologies of modern artificial intelligence and software development and explore the new methodology of organization, stoage, sharing, integration, and management for crop models. The purpose of this study was to adopt the multidisciplinary theories and technologies for studying crop growth system, analyze and synthesize the modeling process, model structure, basic algorithm, parameter type and simulation data for crop growth model on the basis of the concept of crop model resources and integrated model construction, establish the key technology of organization, integration, sharing and management for crop model resources, and construct the Crop Model Resources Warehouse (CMRW) and resource-based Integrated Crop Modeling Construction System (ICMCS), which would lay a foundation for realizing intelligent and automated crop system simulation.Bringing the idea of resource-based integrated manufacturing into the field of crop modeling, the concepts, features and contents for Crop Model Resource (CRM), CMRW and Integrated CMR Construction (ICMRC) were developed. The crop model resources are composed of crop modeling technology resources and software development resources, and have manageable, reproductive, reusable and migration characteristics. The crop system simulation is the dynamic transformation process in the spaces of different states, the integrated construction for CMR provides the technical support for CRM’s formation, and promotes the automatic access, dynamic storage and transfer for modeling technology resources and model software development resources. Development of integrated crop modeling system with the features of resource integration, self-learning, self-adaptation and component design can help researchers to construct and accumulate the unified, shareable and reusable CMR in non-programming environment, reduce the difficulty of model construction and development, and improve the efficiency and quality of the models.Adopting the ontology-based method into the field of crop system simulation, following the main line of crop modeling, the growth models and simulation systems for wheat, rice and cotton crops were comprehensively analyzed, crops simulation and crop model ontology were established, and then Crop System Simulation Framework composed of inner and outer knowledge frameworks for crop modeling and crop model was constructed. Crop simulation model and simulation ontology are the basis of crop simulation ontology, outer knowledge framework for crop modeling can be formed with the system simulation activity ontology as the main line and crops, data, simulation and simulation tasks, and so on as mutual integration ontology. The models or sub-models inner knowledge framework of crop development, biomass production, organogenesis, partitioning and yield formation, water and nutrient balance and so on, can be constructed based on crop growth rules and simulation model components. The crop system simulation framework with a comprehensive knowledge expression promotes the sharing and reusing values of crop model, and thus the design of reusable integrated modeling software architecture.Based on the crop system simulation framework and combining with model representation and component method, the mapping mechanisms of CMR were proposed, the crop growth meta-model and metadata constructed, and the integrated model construction theory and technology established. The crop model can be mapped into the synthesis of concepts and/or trees, model algorithms and/or charts and realization forms for crop model, and the model resources can be expressed as the new entity of "framework + rules + algorithm + template + components". Based on the inner knowledge framework for crop models and sub-models, software patterns and comprehensive data, the crop growth meta-models and metadata were constructed. With the techniques of component framework, description script of model, reflective component assembly, dynamic knowledge integration and model extraction, the CMR integration mechanism can realize the comprehensive synthesis of model structure, model components, algorithm knowledge and model parameters. The crop modeling activity is then transformed to the learning and practicse activity performed together by crop growth meta-model and meta-data, modeling learning mechanism, simulation support tools, model custom component and model engine. The technology of CMR mapping and integrated construction has realized the reuse of crop modeling knowledge and components framework.With the technology of relation database, the crop models were mapped to E-R model, in which crop model structure was constructed by using virtual node method, and modeling knowledge was organized by the integration of knowledge page, knowledge group and algorithm unit, and the properties of model components were described by using faceted classification scheme. By coupling with basal database and solution database for crop models, the relation database model of CMRW was constructed, including model framework library, algorithm library, experiment database and solution database. With the model solution library as the main line, CMRW implements retrieval of model framework, algorithm and parameters by using retrieval algorithm with three layers for CMR. Then, on the integrated database platform, the CMRW for wheat was constructed, thus validating the feasibility of construction and retrieval. The CMRW can accumulate basic resources for integration and reuse of crop model componentsUnder the support of the CMR integrated construction technology and the CMRW, the resource-based ICMCS was successfully developed using .Net platform, language C# and Microsoft SQL Server database, and actually applied to the modeling practices with wheat and barley. ICMCS has the characteristics of hierarchy, components and intelligence, is composed of sub-model frameworks, models and templates for growth period, biomass production, organ production, yield formation, water and nutrient balance. Based on logical crop growth models and modeling service agencies, the functions such as model construction, simulation operation, model preservation and output, model analysis and evaluation, model resource search, and model dictionary and tools are realized. Application tests showed that the ICMCS has strong versatility, adaptability and scalability, and may replace the traditional crop model construction mode of code-based programming. Through using the means of learning, customization, design, configuration, assembly, assessment, expansion, and others, agronomists can intelligently construct, develop, accumulate and test the CMR and preliminary model without programming.更多还原