【作者】 袁庆霓；

【导师】 谢庆生； 许明恒；

【作者基本信息】 西南交通大学 ， 机械制造及其自动化， 2010， 博士

【摘要】 在网络化制造环境下,企业关注的核心是资源信息,因而企业间的业务协同过程,可以看作是企业间优势制造资源的配置和资源重组过程。这个过程涉及了资源的建模、资源发现、资源获取、资源搜索等一系列对资源的操作。合理有效地管理好这些资源及对资源的操作,促进制造资源有效共享,就成为网络化制造环境下制造资源共享服务的核心任务。本文结合语义Web (Semantic Web),构建了基于语义Web的制造资源共享服务平台,对网络化制造环境下的制造资源语义建模、基于本体学习的制造资源自动获取技术、基于概念网的语义匹配、基于规则的本体推理等资源共享服务中的语义技术问题进行深入研究,为充分、合理地共享和利用现有的制造资源信息,实现资源的高效共享,探讨了新的途径,满足网络化制造发展的需求。其主要研究成果如下：(1)综述了制造资源共享的国内外研究现状,针对存在的问题,提出了结合语义Web技术,构建制造资源共享服务平台,并对平台中的语义技术进行了分析,确定了论文的研究内容和思路。(2)针对互联网上不同的企业和组织对资源的概念描述存在的差异,提出一种基于概念层次的语义建模方法。该建模方法引入制造资源的元数据,充分利用现有的资源描述规范和标准,构成了由概念层和元数据层组成的制造资源本体的语义元数据模型。实现了对制造资源的有效组织和描述,保证制造资源语义的一致性和完整性。(3)针对互联网中的制造资源知识信息主要数据格式,以Web表格形式为主要对象,提出了基于Web表格的资源自动获取本体学习方法。该方法利用Spider进行网络主题搜索,经过网页去噪、结构化和文档转换等网页处理后,对网页的元数据信息进行表格内容抽取,并通过一定的规则将提取的概念及概念间的关系自动映射为制造资源领域本体,实现本体库的自动构建。(4)为了提高用户检索条件和制造资源信息的匹配效率,首先结合概念网络与制造资源领域本体,建立了加权概念网,在基于本体加权概念网的基础上,提出了独立元素相似度算法和概念网结构相似度算法相结合的匹配方法。该方法将用户检索条件用基于语言的方法进行概念提取与处理,计算与制造资源领域本体概念的相似度,以相似度最高的本体概念为概念网结构的入口,计算概念网结构相似度。概念网结构相似度采用语义距离来度量,用改进后的最短路径算法来计算语义距离,并构建了相似度函数。(5)针对构建制造资源本体的逻辑冲突问题,确定了基于描述逻辑的领域知识检错(一致性检测)推理和基于关系的蕴涵知识发现推理两种基本推理机制,提出了基于描述逻辑与规则的逻辑推理模型。在该模型中,本体库中的一致性问题,采用基于Tableaux算法的RACER推理实现；在蕴涵知识发现中,在对制造资源领域本体的规则进行定义的基础上,采用SWRL规则语言对规则进行形式化描述,并进行格式的转换,实现基于Rete算法的Jess推理。经过本体的自动逻辑推理,优化了本体结构,为用户界面的本体匹配或检索提供具有合理逻辑的知识层次结构。更多还原

【Abstract】 In networked manufacturing environment, the enterprises pay close attention to the information of resource. The business process between enterprises can be regarded as a configuration and reorganization process of advantage manufacturing resources in cooperative enterprises. This process involves the resources operation including resources modeling, resources discovery, resources obtaining and resource searching and so on. To effectively manage these resources and these operations of resources, and promote manufacturing resources sharing, it becomes the core task of manufacturing resources sharing service in network manufacturing environment.The manufacturing resources sharing service platform based on the Semantic Web is designed. The semantic technology problems in resource sharing service are studied in detail in this thesis. The problems include the semantic modeling of manufacturing resource in the networked manufacturing environment, the automatic acquisition technology based on ontology, the semantic matching based on concept network and the logic reasoning based on rules. New ways are explored in order to the share and use the resource information of manufacture efficiently and resonablely, in order to obtain the efficient source sharing and to meet the needs of network manufacturing development.The following results are achieved in this thesis:(1) The research present status of manufacture source sharing home and abroad is introduced, and aimed at open question the platform of manufacture source sharing service combining with Semantic Web technology is established. The semantic technology in the platform is analyzed and the contents of the thesis and research route are determined.(2) Semantic modeling of manufacturing resources. According to the different concept of resource description of enterprises and organizations on the Internet, semantic modeling method of manufacturing resource based on the conceptual level is proposed. Manufacturing resources metadata models consist of the concept layer and metadata layer are structured by means of manufacturing resources metadata and by making full use of the existing resources description specifications and standards. So that, manufacturing resources can efficiently organized and described, and the consistency and the integrity of manufacturing resources semantic can be ensured.(3) Aimed at the main data format of manufacturing resources in internet and the main object in Web table type,ontology studying method of automatic acquisition source technology based on Web format is proposed. According to the data format of manufacturing resources knowledge on the Internet, the ontology studying method of the automatic acquisition resources based on Web form is put forward. The network resources are searched by Spider in this method. And then, the web page metadata information of table contents is extracted by means of Web page noise erasing, structuring and the documents conversion. The domain ontology is constructed automatically by mapping the concepts and the relationship among the concepts into ontology in the field of manufacturing resources.(4)In order to improve information retrieval efficiency between manufacturing resource and user retrieval conditions, weighted concept network is established firstly. Based on the weighted concept network of ontology, matching method that combines independent element similarity algorithm and structure similarity algorithm of concept network is put forward. This method is used to extract and process the concepts of user’s retrieval condition based on language method, and used to calculate the concept similarity between the concepts and the manufacturing resources domain ontology. The ontology concept with highest similarity is as the entrance of concept network structure to calculate the similarity of concept structure network. The similarity is measured by the semantic distance that is calculated by the shortest path algorithm which is improved ahead, and the semantic similarity functions are constructed finally.(5) Aimed at the logic conflict problems in constructing manufacturing resource ontology, two basic reasoning mechanisms are determined based on field knowledge error detecting of description logic and based on implicit knowledge discovery reasoning of relationship. Then the reasoning model based on description logic and rules is proposed. The consistency problem in ontology is worked out by RACER reasoning based on Tableaux algorithm. In the implication knowledge discovery, based on the definition of manufacturing resources domain ontology rule, the formalization description of rules is carried out by SWRL rule language and the rule format is transformed, and then Jess reasoning based on the Rete algorithm is implemented. The ontology structure is optimized by means of automatic logic reasoning of ontology. And the knowledge level structure with reasonable logic is supplied for the ontology matching or retrieval of user’s interface.更多还原