节点文献
铁路信息共享的Web服务组合关键技术研究
Research on Key Technologies of Web Service Composition in Railway Information Sharing
【作者】 马小宁;
【导师】 董宝田;
【作者基本信息】 北京交通大学 , 交通运输规划与管理, 2010, 博士
【摘要】 经过多年的努力,我国铁路信息化建设取得了一定成绩,建成了若干铁路业务信息系统,但是,各业务信息系统大多相互独立、自成体系,没有构成一个有机整体,导致各信息系统间信息交换困难,信息资源难以共享,综合利用难以展开,整体效益难以发挥,严重制约了铁路信息化的发展,其现状已不能适应铁路运输快速发展的需要。面向服务架构(SOA)为突破信息孤岛、整合信息资源、深化业务协同、缩短开发周期、降低开发成本的铁路信息化应用带来了新的曙光。SOA的基本思想是以服务为核心,首先将企业的IT资源整合成可操作的、基于标准的服务,进而通过服务的快速组合和重用,保证应用的敏捷性与扩展性,满足铁路信息化发展的需要。基于SOA及Web服务组合的架构对于实现铁路信息共享具有重大的意义。本文以铁路信息共享的需求为背景,对基于SOA的铁路信息共享及Web服务组合关键技术进行研究。首先在对铁路信息共享需求分析的基础上,提出了基于SOA的铁路信息共享平台架构,分析了铁路信息共享的策略及实施要点;针对现有Web服务组合方法存在的局限性,提出了基于与/或树的Web服务组合模型,对算法进行了程序实现,并结合算例验证了模型的合理性及算法的有效性;将线性物理规划理论引入到Web服务选择中,在对其进行改进的基础上,建立了基于线性物理规划及遗传算法的Web服务选择模型,对其进行了程序实现,并通过算例证明了方法的可行性及适用性。本文的主要研究内容及成果包括如下几个方面:(1)基于SOA的铁路信息共享平台针对铁路信息化建设中存在的信息孤岛问题,分析了铁路信息共享的需求,提出了基于SOA的铁路信息共享平台技术架构,分析了铁路信息共享的策略及实施要点。(2)基于与/或树的Web服务组合分析了基于智能规划及图搜索的Web服务组合方法在输入/输出依赖关系的表达上所存在的局限性,提出了基于与/或树的Web服务组合模型。首先给出了输入/输出依赖与/或树的形式化定义,进而讨论了构建输入/输出依赖与/或树的搜索算法,包括基本的与/或树搜索算法及针对Web服务组合的搜索策略。基于卡内基梅隆大学的“C++搜索类库”对Web服务组合的与/或树搜索算法进行了实现,并以铁路货物运输中货物运输费用的计算流程为例,分析了基于与/或树的Web服务组合的实施步骤,验证了模型的合理性及算法的有效性。(3)对线性物理规划的两点改进分析了线性物理规划中存在的两点局限性:求解凸度系数时存在的反复迭代及偏好函数表达形式的复杂性,提出了无循环的求解凸度系数的方法及针对Web服务选择问题的形式简单的分段线性偏好函数。(4)基于线性物理规划及遗传算法的Web服务选择针对现有的基于简单加权法的Web服务选择所存在的局限性,提出了基于线性物理规划及遗传算法的Web服务选择模型,包括基于线性物理规划的组合服务QoS评价方法以及基于遗传算法的Web服务选择求解模型。首先给出了该方法的实施步骤,进而对基于遗传算法的Web服务选择进行了程序实现,并通过算例证明了方法的可行性及适用性。本文以铁路信息共享的需求为背景,对基于SOA的铁路信息共享及Web服务组合中的关键问题进行了研究,对于突破信息孤岛,实现信息共享及服务之间的无缝集成,形成功能强大的企业级流程服务以完成企业的业务目标提供了一定的理论支持。
【Abstract】 After many years of hard work, we have gotten certain achievement in railway informatization construction. Many railway business information systems has been built, but these information systems are mostly independent and self-existent, and did not construct a organic whole, result in that information exchange and sharing is difficult, comprehensive utilization is difficult to deploy, whole benefit is difficult to play, and this seriously restricts the development of railway informatization construction, present situation can not adapt rapid development requirement of railway transportation.Service-Oriented Architecture (SOA) bring new dawning for railway informatization application to break through information island, integrate information resources, shorten development period and reduce development cost. The basic idea of SOA is taking service as core, integrating enterprise IT resources into operable and standards-based services, through rapid composition and reuse to ensure agility and expansibility of application, and to meet the requirement of railway informatization development. SOA and web service composition based architecture has great meanings to realize railway information sharing.This paper takes the requirements of railway information sharing as background, and studied SOA based railway information sharing and web service composition. Firstly based on the requirement analysis of railway information sharing, SOA based railway information sharing platform architecture is proposed; according to the limitation of existing web service composition method, AND/OR tree-based approach is proposed with program realization, and verified rationality of the model and effectiveness of the algorithm with numerical example; introduced Linear Physical Programming into web service selection field, based on the improvement of it, established Linear Physical Programming and Genetic Algorithm based web service selection model with program realization, and explained implementation details with numerical example. Main study contents and results of this paper include following several aspects:(1) SOA based Railway Information Sharing PlatformAccording to the problems of Information Island in railway informatization construction, analyzed the requirement of railway information sharing, and proposed SOA based railway information sharing platform architecture, then emphatically described solution and implementation details of railway information sharing.(2) AND/OR Tree-based Approach for Web Service CompositionAnalyzed the description limitation of input/output dependency relationship in AI planning and graph searching based web service composition, proposed AND/OR tree-based approach for web service composition. Firstly formal definition of Input/Output Dependency AND/OR Tree is proposed, furthermore described searching algorithm to built Input/Output Dependency AND/OR Tree, including basic AND/OR tree search algorithm and special algorithm. We realized AND/OR tree search algorithm based on "AI C++ Search Class Library", and took the freight transportation cost calculation flow as example to explain implementation steps of AND/OR tree-based approach for web service composition.(3) Tow Improvement of Linear Physical ProgrammingAnalyzed two limitations of Linear Physical Programming:repeated iteration when solve convexity parameter and complex form of preference function, proposed new method to determine proper initial value of convexity parameter without iterations and simple form of preference function.(4) Linear Physical Programming and Genetic Algorithm based Web Service SelectionAccording to the limitation of existing weight-based web service selection, Linear Physical Programming and Genetic Algorithm based web service selection model is proposed, including Linear Physical Programming based evaluation method of composite service QoS and Genetic Algorithm based web service selection model. Firstly execution steps of this method are proposed, furthermore realized it in program, and explained implementation details with numerical example.This paper takes the requirements of railway information sharing as background, and studied SOA based railway information sharing and web service composition. This has theoretical support to break through Information Island, realize information sharing and seamless integration between web services, form powerful enterprise business flow to accomplish enterprise business target.
【Key words】 information sharing; SOA; Web service; Web service composition; AND/OR tree; QoS; Physical Programming; Genetic Algorithm;