【作者】 蒋蕊；

【导师】 张仲义；

【作者基本信息】 北京交通大学 ， 系统工程， 2010， 博士

【摘要】 本文将网格技术应用到企业信息系统集成领域,着重对网格环境下企业信息系统集成(Enterprise Information System Integration Based on Grid, EISIBG)的体系结构框架、模型实现、平台设计、构建方法与资源调度策略等几个关键问题进行研究,以期有效地消除“信息孤岛”,更加高效、便捷地访问所需的各种数据。论文主要工作如下：首先,概括网格的基本理论,包括网格的概念、体系结构和多智能体理论,分析企业信息共享与集成的研究现状以及企业信息系统集成在网格时代面临的机遇和挑战,探讨EISIBG的选题背景和研究意义。其次,针对企业信息系统特点,采用网格层次化设计思想,对OGSA网格体系结构进行改进和扩展,提出一种新的易于使用网格服务的EISIBG层次化体系结构；在此基础上设计基于FEAF的六层EISIBG体系结构框架,使用视角矩阵图展示体系结构框架不同部分的各个方面,并分析EISIBG的原型系统模型实现,以及基于UML设计EISIBG体系结构框架产品的模型实现。第三,针对目前企业信息系统中普遍存在的数据和业务异构问题,引入多智能体系统(Multi-Agents System, MAS),提出多智能体EISIBG集成平台。分析平台的Agent组成及功能,设计平台的基本构成单元——服务驱动型协作Agent的元模型,在此基础上,研究EISIBG平台MAS联盟的协作模式,给出MAS联盟的形式化语义,讨论MAS联盟的服务流、信息流和断点工作流设计,以联盟的方式提高MAS的全局效率,为企业信息系统网格环境集成提供有力的平台支持,最后,佐证说明。第四,提出一种网格环境下企业信息系统构建方法,研究EISIBG构件的定义、知识库的组成和划分,并深入探讨各类构件的设计和生产方法,最后提出基于按单定制思想的企业信息系统组装和实现过程。第五,根据调度任务是否存在前后依赖关系,分别建立独立型任务调度模型和依赖型任务调度模型。在此基础上,针对独立型任务调度提出一种改进算法,算法着重考虑特定任务只能由特定资源来完成的问题,从而引入资源计算能力方面的约束；针对依赖型任务调度提出一种新的任务调度策略,着重考虑各子任务执行时的前后次序,最后,通过仿真验证所提算法的有效性。更多还原

【Abstract】 This dissertation applies the Grid Technology to the area of Enterprise Information System Integration (EISI), focusing on several key issues of Enterprise Information System Integration Based on Grid (EISIBG) such as the architecture framework, model implementation, platform design, construction methods and resource scheduling strategy. The target is to effectively eliminate "Information Island" and to access variety of data more efficient and convenient. The main contributions are as below:First, the fundamental theory of Grid is summarized, including Grid concept, architecture and Multi-Agents theory. The current research on enterprise information sharing and integration and the opportunities and challenges of EISI at Grid times are analyzed. The research background and significance of EISIBG were also addressed.Second, based on the characteristics of enterprise information system, a new EISIBG hierarchical system architecture easy to apply Grid services is proposed by adopting hierarchical grid design idea and improving and expanding OGSA grid architecture. In addition, FEAF-based six-layer EISIBG architecture framework is designed and various aspects of different parts of the framework are displayed using the architecture perspective matrix diagram. Furthermore, EISIBG prototype system modeling is investigated and EISIBG architecture framework products model based on UML is developed.Third, to solve the common issues of data sources and service heterogeneous in current enterprise information system, Multi-Agents System (MAS) and Multi-Agents EISIBG integration platform are introduced. Through the analysis of Agent composition and function of the platform, the basic component of EISIBG platform, the service-driven collaborated Agent meta-model, is built. Based on the above models, MAS Union collaboration mode on EISIBG platform is further studied and its formal semantics is provided. The service flow, information flow and breakpoint workflow of MAS Union are designed to improve the overall efficiency of MAS and to provide the strong platform support for EISIBG integration. Finally, some examples are provided.Fourth, an enterprise information system building technique under grid is proposed. Then EISIBG component definition, knowledge library composition and division, and the design and production methods of different components are further investigated. Finally, the integration and implementation process of enterprise information system based on Build-To-Order is presented.Fifth, according to the dependency between the scheduling tasks, the Independent Task Scheduling model and the Dependent Task Scheduling model are built individually. For Independent Task Scheduling, an improved algorithm is proposed which focuses on the feature that a specific task can only be completed by a specific resource and thus to bring in the resources computing capacity constraints. For Dependent Task Scheduling, a new scheduling strategy focusing on the execution order of sub-tasks is proposed. Finally, the effectiveness of the algorithm is verified by simulation.更多还原