中国科学院机构知识库网格
Chinese Academy of Sciences Institutional Repositories Grid
基于过程数据的企业集成关键技术研究与应用

文献类型:学位论文

作者刘伟
学位类别博士
答辩日期2008-06-03
授予单位中国科学院研究生院
授予地点中国科学院软件研究所
导师戴国忠
关键词过程数据 企业集成 发布/订阅 QoS ECA 数据融合
其他题名Research on Key Issues of Enterprise Integration Based on Process Data
学位专业计算机应用技术
中文摘要流程企业综合自动化是实时生产管理集成优化的核心,而综合自动化的基础是企业生产过程数据的有效集成。企业生产过程数据主要包括生产运行与管理涉及到的实时和历史数据、事件、消息等。与传统企业集成不同的是过程数据具有不同的时间周期、不同的概念外延、生产工艺知识约束以及实时性要求。此外,企业集成环境的复杂性,如传感器数据高噪音、异步采样等也增加了过程数据集成的难度。论文以流程企业过程数据集成为背景,研究模型驱动的过程数据集成技术,重点研究过程数据集成模型、模型驱动的过程数据集成集成框架、QoS自适应的实时发布订阅机制以及基于反馈的多传感器数据融合算法,力图研发一个支持不同尺度的过程数据集成的工具。论文首先分析流程企业生产过程数据集成的特点和传统企业数据集成建模方法的不足,提出了一种基于模型驱动的企业过程数据集成方法,采用领域本体的方法从时态对象、集成过程、语义集成三个角度建立过程数据集成模型,并对这三个模型进行了形式化描述,此外,通过定义映射规则,实现应用本体间关系的映射。在分析分布式事件通知服务体系结构的基础上,提出了基于发布订阅的过程数据集成框架,通过采用事件-条件-动作(ECA规则)来支持模型驱动的企业过程数据集成方法。为了简化过程数据集成模型的建模,提出了一种可视化的ECA规则描述规范,并设计开发了相应的编译工具。针对传统的分布式事件处理不能满足流程企业中过程数据集成的实时性要求和QoS保障的问题,论文在传统分布式事件处理之上扩展设计了QoS保障策略和带截止期的ECA规则(RECA),提出了一种自适应发布订阅的机制,该机制通过动态调整系统参数,可以同时提供多层次的服务质量。实验数据验证了该机制能够提高多服务请求并发情况下,不同QoS等级的响应处理和可预测性。进一步,针对企业过程数据集成中面临的传感器采集高噪音、异步采样等问题,论文给出了多传感器数据融合的一种数学描述,在比较分析典型的多传感器数据融合的算法的基础上,提出了一种基于反馈控制原理的多传感器数据融合算法,并给出了算法实现及实验数据验证。最后,基于上述研究成果,论文设计并开发实现了一个适应大规模分布式流程企业生产过程数据集成的自适应实时发布订阅服务系统,并作为流程企业生产执行系统(SMES)的核心构件,在多家石化企业得到了成功应用。
索取号暂无
英文摘要The core of production optimization in real-time is the Process Automation. It is imperative to integrate production process data with different periods. Usually, production process data include real-time/historical data, messages and events etc. Different from traditional enterprise data integration, process data integration is challenged by the different time-cycles, the different extension concept of data, the constraint of production process knowledge and the requirement of real-time guarantee. Moreover, the complexities of enterprise integration environment, such as high noise in sensor data, asynchronous sampling etc, make process data integration more difficult. In the context of process data integration of Process Industry, the author focuses on model-driven process data integration, specially on the model of process data integration, the integration architecture of model-driven process data integration, QoS adaptive real-time publish-subscribe mechanisms, as well as multi-sensor fusion algorithm based on feedback, aiming to develop a tool which can support process data integration with multi-scale. Based on the analysis of process data feature and the inadequacies of traditional enterprise integration modeling, the author presents a model-driven data integration method, using domain ontology in the perspective of temporal objects, integrated process, and semantic integration to build the model of process data, and gives formal description for these three models. Besides, the relationship mapping between application ontology is given. Moreover, through the event-condition-action (ECA rule), an architecture which is based on publish/subscribe is presented for process data integration to support the enterprise model-driven process data integration methods. To simplify the process of data integration modeling, a visual ECA specification and the complier are given. In addition, to solve the problems of real time data integration and the guarantee of QoS, the author develops a guarantee strategy of QoS and ECA rules (RECA) with the deadline. Adaptive subscription and publish service is implemented to support the strategy, which can provide service quality for multiple levels through dynamic adjustment of system parameters. Experiments show that the mechanism can improve the processing response of different QoS levels and predictability in the complicated circumstances with multiple service requests. Further more, to overcome the problem of high noise in sensor sampling in process data integration, the author gives a mathematical description of multiple data sources fusion in Process Enterprise. And based on the comparison of the typical multi-source data integration algorithms, a new feedback control algorithms is proposed which based on AFASRNT. Finally, based on the research results, we design and develop a large-scale distributed and real-time adaptive publish-subscribe service system which is used as the key component in manufacturing execution system of SINOPEC,and has been validated in quite a lot petrochemical enterprises.
公开日期2011-03-17
分类号暂无
源URL[http://124.16.136.157/handle/311060/7054]  
专题软件研究所_人机交互技术与智能信息处理实验室_学位论文
推荐引用方式
GB/T 7714
刘伟. 基于过程数据的企业集成关键技术研究与应用[D]. 中国科学院软件研究所. 中国科学院研究生院. 2008.

入库方式: OAI收割

来源:软件研究所

浏览0
下载0
收藏0
其他版本

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。