地理知识云服务发现与组合技术研究

发布时间：2018-02-27 23:35

本文关键词： 地理知识云云服务地理决策空间数据挖掘服务描述服务发现服务组合蚁群算法流程引擎地震影响场分析　出处：《福州大学》2014年博士论文　论文类型：学位论文

【摘要】：人类已快速进入前所未有的大数据时代,人类知识的获取和应用模式正从注重个体创新走向强调群体共享和协同聚集的新阶段,全球范围内日益增加、分布存储的各类数据挖掘与知识发现算法、决策分析模型只有通过协同地提供应用才能发挥其应有的价值。地理知识云服务平台是当前空间数据挖掘、地理决策分析、知识共享服务领域最富有前途和最具挑战性的前沿课题之一,是现代信息服务业未来发展的核心技术,可为我国空间信息服务业的跨越发展、产业升级提供共性技术与支撑平台。本文以地理知识云服务的深度共享和分布式协同应用为目标,综合运用分布式空间数据挖掘与地理决策、服务式计算、云计算等先进技术和理念,从知识云服务的描述、发布、发现、动态组合与优化、分布式执行等方面,对地理知识云服务的管理和应用展开了系统性的研究。主要研究内容和成果如下：(1)研究与总结了地理知识云服务的概念和特征,创新性地提出了广义和狭义地理知识云的具体定义,建立了一个融合通用指标和领域指标的地理知识云服务QoS质量模型,提出了一种基于WSDL扩展的云服务QoS描述方法；研究了基于文本、接口和语义等的服务发现技术,根据地理知识云服务需要处理大数据的要求,提出了一种基于数据格式的服务匹配方法。(2)提出了一个基于Hadoop框架的地理知识云服务注册中心和基于MapReduce的并行服务发现框架。该中心面向列存储模式,建立了基于服务类别和功能的双重索引结构,并基于HBase数据库实现了知识云服务描述信息的分布式存储。中心可方便地实现存储系统的横向扩展,解决了传统服务注册中心的访问性能低下和多节点数据同步较为复杂的问题,同时提供了快速的并行化服务发现手段。(3)研究了地理知识云服务组合的多目标优化问题,提出一个以蚁群算法为基础且考虑数据传输代价的多目标服务组合优化算法DVA-MOACO。该算法建立了个考虑多QoS指标的服务质量评估模型,通过改进蚁群的状态转移概率计算方法,使蚁群寻径时同时考虑数据传输代价和QoS质量。为加快算法的搜索过程,根据蚁群算法的并行特性和云的并行计算条件,提出了算法的一个并行化版本。该算法可快速地从多个组合服务的备选方案中找到整体QoS性能较优、数据传输代价较低的Pareto最优解,以满足知识云服务在线应用时高质量和即时性的要求。(4)设计了一个分布式地理知识云服务组合软件引擎,从流程模型、体系结构、流程解析、分布式执行和异常处理等方面对引擎进行设计,提出了图形-模型一体化的流程模型和四层式的组合服务执行引擎体系架构,实现了主从模式和P2P模式相结合的组合服务分布式执行方法,并提供了基于日志的异常处理功能。引擎充分利用云平台提供的多虚拟机环境,实现了地理知识云组合服务的高效、可靠运行。(5)以福建地区为例,开展了地理知识云服务在GeoKS-Cloud平台上提供地震影响场分析服务的具体应用。从知识服务的分解与部署、组合服务建模、服务执行等方面呈现了地理知识云服务的分布式、协同服务过程。应用结果表明,所构建的地震影响场分析云服务,其计算得到的地表地震动峰值加速度的分布总体符合椭圆型衰减规律且根据不同场地类别进行了校正,云服务与传统的网络服务相比具有可扩展、可配置、开放透明等明显优势,不仅可提高分析运算的效率,还能支持数据和模型算法等资源的复用。
[Abstract]:Human beings have entered the era of big data acquisition and application mode of hitherto unknown, human knowledge is from the emphasis on individual innovation to highlight the new stage of group sharing and collaborative aggregation, globally distributed storage is increasing, all kinds of data mining and knowledge discovery algorithm, decision analysis model only through collaborative applications to play its due the value of geographical knowledge. The cloud service platform is the geographic spatial data mining, decision analysis, knowledge sharing service field is the most promising and challenging topic in the future is the core technology of the development of modern information service industry, for the spatial information service industry in our country the leapfrog development, industrial upgrading and provide common technology support platform. According to the geographic knowledge service cloud depth sharing and distributed collaborative application as the goal, the comprehensive use of distributed spatial data mining And geographic decision-making, service computing, advanced technologies such as cloud computing and cloud service concept, from the knowledge description, publishing, discovery, dynamic combination and optimization, distributed implementation, management and application of geographical knowledge of cloud services carried out a systematic research. The main research contents and results are as follows: (1) study and summarizes the concept and characteristics of geographical knowledge of cloud services, put forward the innovation of the broad and narrow geographical knowledge of the specific definition of cloud, a fusion of common indicators and indicators in the field of geographic knowledge QoS cloud service quality model, proposed a method to describe the extension of WSDL cloud service based on QoS; based on text, interface and semantic service discovery technology, according to the geographical knowledge of cloud services need to handle large data requirements, a method is proposed for matching data format based on the service. (2) proposed a framework based on Hadoop The geographical knowledge of cloud service registry and parallel service discovery based on MapReduce framework. The center column oriented storage mode, set up a dual index structure based on service types and functions, and distributed storage HBase database knowledge cloud service description based on the information center can conveniently realize the horizontal expansion of storage system, solve the access the poor performance of traditional service register center and multi node data synchronization is quite complex, while providing a parallel service rapid discovery method. (3) the multi-objective optimization problem of geographical knowledge cloud service composition, proposed an ant colony algorithm based and consider the data transmission cost of multi-objective optimization algorithm the DVA-MOACO. algorithm is established considering multi QoS index service quality evaluation model, the improved ant colony state transition probability calculation method, the The ant colony routing considering the data transmission cost and the quality of QoS. In order to speed up the search process, according to the calculation conditions of parallel ant colony algorithm and parallel characteristics of cloud, we propose a parallel version of the algorithm. The algorithm can quickly from multiple combinations of service alternatives in the whole performance of QoS is better. Pareto optimal data transmission with lower cost solutions to meet the knowledge of cloud services online application of high quality and real-time requirements. (4) the design of a distributed geographic knowledge cloud service composition software engine from the process model, system structure, flow analysis, distributed execution and exception handling design of engine and put forward the process graphical model integration and four layer composite service execution engine architecture, and implement a method of composite service with distributed master-slave mode and P2P model, and provide The exception handling function based on log. Virtual machine environment engine takes full advantage of cloud platform, realizes the combination of geographical knowledge cloud service efficient, reliable operation. (5) in Fujian area as an example, the geographical knowledge of cloud services to provide earthquake influence field analysis of specific application services on the GeoKS-Cloud platform. The decomposition of knowledge service and deployment, combination of service modeling, service execution and other aspects presents a distributed geographic knowledge cloud services, collaborative service process. The application results show that the cloud service analysis of seismic influence field constructed by the overall distribution, the peak acceleration of the surface to meet the calculation of elliptic attenuation law and according to different site categories the correction, compared with the traditional network service cloud service with scalable, configurable, open and transparent and other obvious advantages, not only can improve the analysis operation efficiency, but also support Reuse of resources such as data and model algorithms.

【学位授予单位】：福州大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TP393.09;P315

【相似文献】