【作者】 吴雷；

【导师】 魏臻；

【作者基本信息】 合肥工业大学 ， 计算机应用技术， 2012， 博士

【摘要】 传统的ERP系统是在企业应用与管理策略驱动下的多功能集成应用系统,其结构相对复杂、封闭且灵活性较差,无法适应现代服务业企业动态多变的业务流程,也无法满足现代服务业高度协同和异构分布的复杂特性。因此,如何提高ERP系统的开发、维护效率,提高系统的灵活性与开放性,并能支持跨企业间动态协作,便成为一个有价值的研究课题。面向服务架构下的模型驱动开发方法,为现代服务业的ERP系统开发提供了一种新的企业模型驱动的应用开发体系。本文以某大型出版传媒企业ERP项目为依托,结合现代服务业企业的实际需求,以企业模型驱动的ERP开发方法和ERP系统中的优化算法研究为主线,系统研究了现代服务业ERP开发中的企业建模、模型转换、构件识别、优化算法等若干关键技术问题,提出了一个面向服务的ERP轻量级开发框架,为现代服务业的ERP应用产品提供了一种快速实施方法,并为今后ERP功能扩展时需要的大规模软件集成与重用提供了解决方案。本文的主要研究内容如下：(1)提出了一套ERP系统快速开发与实施框架。将模型驱动开发方法学和面向服务架构融入到ERP系统开发过程中,建立了一套轻量级的ERP系统快速开发框架：模型驱动方法的应用使得用户需求和业务的变化可以及时反映到应用系统代码中,从而节省软件的开发成本以及维护工作量,提高开发效率；面向服务架构的应用增强了ERP系统的开放性和可扩展性,使得企业内部、供应链上下游企业之间的应用系统交互成为了可能,为实现SOA下软件的低耦合集成与重构提供了解决方案。(2)在模型驱动架构的基础上,将元建模理论应用于ERP领域建模,建立了面向现代服务业ERP开发的领域模型。对BPMN的元模型进行改进,扩展出了一个以过程模型为核心,涵盖组织、功能、信息、资源四个方面的ERP业务领域元模型,并对构成元模型的各元素以及各元素之间的关系进行了描述；其次,借鉴MOF中对元模型的描述方法,采用XMLSchema对所提元模型进行了基于XML的描述。上述工作也为企业模型向下一级别的系统模型转换奠定了建模基础。(3)提出了一种从企业业务模型向Web Services架构下的企业应用模型转换的方法。首先建立了面向服务架构下的服务模型,并对服务模型的元模型构成作了形式化说明；其次,采用基于元模型的模型转换理论,阐述了从业务模型向服务模型转换的映射规则与方法。模型驱动的开发方法,将业务逻辑与具体的编码实现进行了有效分离,使得设计人员和编程人员可以关注各自的领域,并能通过对业务模型的重用及可变分析实现应用的快速开发,在一定程度上提高了开发效率。(4)为获取面向服务架构下适合的服务构件,提出了一种从服务模型中获取服务构件的识别方法。在服务构件识别过程中,采用数学方法对服务模型作了形式化描述,通过服务之间的关联度计算公式,形成关联度矩阵,采用聚类分析方法对矩阵进行分析,从而获取高内聚、低耦合、可复用的业务构件,提高了框架的可重用性。(5)通过对粒子群算法进行改进,提出了一种基于Memetic框架的混合粒子群算法。该算法能够在加快收敛速度的同时提高算法的搜索多样性,避免算法陷入局部最优,并通过实验证明了所提算法的有效性。随后研究了该算法在离散型优化领域的应用,采用基于实数的编码映射方案,使得该算法能够应用在现代服务业ERP的相关离散型优化问题上,并在物流配送调度及路径优化方面作了实例验证。更多还原

【Abstract】 A traditional ERP system is a multifunctional integrated application system driven by the enterprise application and management strategy. Because of relatively complicated, closed and poor flexible structure, it can neither adapt to the dynamic and diverse business process, nor meet the demand of highly cooperation and heterogeneous distribution in the modern service industry. How to improve the efficiency of ERP system development and maintenance, and how to enhance flexibility and openness to support dynamic cooperation between enterprises is a valuable research topic. The Model-Driven development method in service-oriented architecture provides a new business model-driven application development method for ERP system development of the modern service industry. Through demand analysis to the modern service industry based on an ERP system of a big publishing and media enterprise, the dissertation systematically study the enterprise modeling, model transforming, component identification, optimization algorithm and so on, with focus on business model-driven ERP development method and optimization algorithm. The dissertation proposes a lightweight ERP development framework in service-oriented architecture which provides a fast implementation method of ERP application products in modern service and offers a solution of the large-scale software integration and reuse for the future. The main contents are as follows:(1) The dissertation provides a fast development and implementation method of ERP system. It combines model-driven development methodology and service-oriented architecture with ERP system development process, and establishes a fast lightweight ERP development framework. By the use of the model-driven development method, the users’ requirements and the changes of business can be timely reflected to the codes of applications. Therefore, the model-driven development method not only saves the cost and maintenance of software development, but also improves the development efficiency. By the use of service-oriented architecture, the openness and extensibility of ERP system are enhanced. It becomes possible that the applications within the internal enterprises and between the supply chain enterprises can be integrated and reused. The service-oriented architecture also gives the solution of service application loose coupling integration and reconstruction.(2) The dissertation establishes ERP development domain model oriented modern service base on the theory of meta-model modeling by applying the model-driven framework to ERP domain modeling. Firstly, the dissertation makes some improvement on the BPMN meta-model, and generates an ERP business domain meta-model whose core is process model. The meta-model covers four aspects including organization, function, information and resource, and describes the relationships between these elements. Secondly, the dissertation makes use of XML Schema to give the XML description of meta-model according to the meta-model description method in MOF. These jobs establish the basic of model conversion from business model to the next level system model.(3) The dissertation proposes a model conversion method from business model to application model under the Web Services framework. Firstly, it establishes a service model in SOA and describes the structure of its meta-model. Secondly, it depicts the mapping rules and method from business model to service model based on meta-model. The model-driven development method effectively separates the business logic and the coding. So that designers and programmers can focus on their own field. And the development method can implement the fast development by reusing the business models and variable analysis, which improves the efficiency of development to a certain extent.(4) In order to obtain proper service components in SOA, the dissertation puts forward a service component identification method from service model. In the process of service component identification, some mathematical methods are used to describe the service model. The correlation matrix is generated by the correlation formula between the services. The high-cohesion, low-coupling and reusable service components are produced by cluster analysis to matrixes, which improves the reusability.(5) A hybrid PSO algorithm based on Memetic framework (HM-PSO) is proposed by improving particle swarm optimization algorithm. While accelerating convergence rate, the proposed algorithm can avoid falling into the local optimum by enhancing the diversity of swarm. The experimental results show that the improved algorithm is an effective optimization algorithm. Then how to apply the algorithm to the discrete optimization domain is been researched. The results show that the algorithm can be applied to relative discrete optimization in the ERP of modern service industry based on the mapping rules in real-coding. And the algorithm is verified effectively in the problem of logistics scheduling and routing optimization.更多还原