【作者】 姜亮；

【导师】 屈福政；

【作者基本信息】 大连理工大学 ， 机械设计及理论， 2009， 博士

【摘要】 随着人工智能和互联网技术不断交叉融合,第三代万维网—语义网及其相关技术正迅猛发展并日渐成熟。语义网技术的主旨在于实现跨平台跨系统的知识共享、知识重用、互操作,并最终实现信息、知识自动处理和应用。当前,生物医疗、地理信息、图书情报等各领域正积极开展语义网技术的应用研究。语义网技术在机械设计领域中的应用研究也已起步并逐渐成为研究热点。然而,目前基于语义网技术的机械设计研究主要集中在设计知识共享、设计知识重用以及跨平台互操作等知识集成方面,对基于语义网技术的机械设计方法和设计信息自动处理等方面却鲜有研究涉足。在国家自然科学基金项目“面向中小企业的基于CAD知识和工程语义的设计知识重用研究”的支持下,本文就语义网技术对机械设计中的创新设计、变型设计、详细设计的支持方式和方法进行了研究,主要工作如下:在对现存本体构建方法进行全面综述的基础上,参考其他领域本体构建方法,总结归纳出一种较全面的领域本体构建方法。其中提出的针对典型应用中间扩展本体建模法,可使本体在任何进化阶段都保证完整地回答一部分能力问题。提出的体现本体建模原则强一致弱耦合的分而治之本体建模法,可解决一次性构建大本体本体承诺过多,不利于重用的难题。并以需求设计本体的构建为例,对此方法进行了实例分析。在创新设计方面,根据技术创新理论TRIZ的重要发现.跨领域重用创新知识和经验可提高技术创新的等级,以TRIZ理论的冲突矩阵为基础,构建了可跨领域统一描述技术创新案例创新本质的语言—技术创新本体,并提出基于本体的技术冲突解决方法,给出基于本体的技术冲突解决流程。此方法可跨领域检索、共享创新本质相似的创新案例,并以相似案例辅助克服传统冲突矩阵的两个应用瓶颈:具体技术冲突标准化和标准解具体化。在变型设计方面,针对基于范例推理方法所面临的主要问题:范例表示和范例修改,提出一种基于领域CBR本体的范例表示和范例修改方法。该方法引入本体表示范例,提高了范例库的共享性、重用性;明确定义范例特征之间的关系,精确了范例的内容意义;本体的一致性检查,确保了范例的有效性。添加范例特征意图项指导范例修正,使范例修正在对范例特征意图明确的基础上进行,可避免源范例原已避开的错误,提高范例修正的成功率。在详细设计方面,共做了两方面探讨:基于隐含知识的技术文档检索和基于本体和MAS的文档知识自动应用。首先,提出“以隐含知识作为检索对象检索出的技术文档同样具有参考价值”的思想,并基于此提出基于本体的技术文档隐含知识检索框架。应用SWRL规则对本体实例知识库中对应技术文档的显式知识进行推理扩展,得出隐含知识作为语义检索的检索对象检索技术文档,提高了语义检索的查全率。其次,借鉴语义网架构思想,提出由计算机基于文档内容含义查找所需文档片段,读懂文档片段所含知识并自动予以应用的基于本体和MAS的机械设计智能文档框架。从标注对象和插入语义信息两个维度给出实现智能文档框架所需的细粒度语义标注的基本特征。接着提出实现智能文档框架必须针对不同知识类型采取不同标注策略的思想。在此基础上,提出了符合FIPA(Foundation for Intelligent Physical Agents)智能Agent参考模型标准的基于本体和MAS的机械设计智能文档框架,并介绍了框架的主要组成部分以及各自功能:一个Agent平台、四个Agent社会以及机械设计本体库。最后对该框架在语义标注、语义检索、文档知识自动应用三个方面进行应用分析。以上述各方法为单元技术,提出了可对创新设计、变型设计、详细设计进行全面支持的基于语义网技术机械设计集成系统框架,并对框架的组成部分和实现方案进行了阐述。以斗轮挖掘机的设计为例,给出了上述各方法的实例分析,验证了语义网技术在机械设计中应用的可行性。最后,总结了全文的主要研究成果及创新点,并给出了未来的研究方向。更多还原

【Abstract】 With the cross merging of artificial intelligence and internet technologies, the Semantic Web technologies, whose aim is realizing the automatic manipulation of information contained in normal Web pages, are developing fast and becoming ripe. At the same time, in many fields like biology, geography, library and information etc., the application studies of Semantic Web technologies are already under way. The application study of Semantic Web technologies in mechanical design field is already start up and becoming research focus. While, at present, research activities are mainly focused on information and knowledge integration, for example, design knowledge sharing, reusing and cross-plateform interoperation etc. Few studies devote themselves into the studies of Semantic Web technologies based mechanical design methods and automatic utilization of design information. Funded by national natural science foundation project named "Study on CAD knowledge and engineering semantic based design knowledge reuse oriented by middle and small enterprise", this thesis performs several key techniques studies on ontology and semantic web based mechanical design.On the basis of summarizing the existent ontology construction methods, by consulting other domain ontology construction methods and by combining personal experience in mechanical domain ontology construction, a domain ontology construction method, which is considered to be more comprehensive, is proposed. With the construction of requirement design ontology as a case, the pheasibility of this method is proved.With regard to the innovative design, by combining ontology and Semantic Web technologies, which aim to realize reuse and sharing of information and knowledge, with theory of inventive problem solving TRIZ, method of solving technical conflict based on ontology is proposed and the flow chart of solving technical conflict based on ontology is given. By making use of ontology compile software Protege and ontology description language OWL, on the base of technical conflict solution theory, which comes from TRIZ, the technical innovation ontology is constructed, which paves way for ontology based technical innovation in semantic web in the future.With regard to the adaptive design, a new case representation and case revision method based on domain CBR ontology is proposed. In this method, the reusability of case base is improved by making use of ontology to represent cases. The successful rate of case revision is raised by adding case feature intention class to domain CBR ontology. Firstly, with classical concept of ontology and CBR theory as foundation, the definition and components of domain CBR ontology are abstracted. Secondly, the superiorities of ontology over traditional way of case representation are given and the key class of domain CBR ontology- case feature intention class and its function are emphatically introduced. Thirdly, the detailed processes of construction of domain CBR ontology and case revision based on domain CBR ontology are given. In the end, domain CBR ontology of bucket wheel, named bucket wheel CBR ontology, is described in OWL.With regard to the detailed design, two aspectes study are performed. First of all, A framework of ontology based technical document tacit knowledge retrieval is proposed. In this framework, by using the SWRL rule to extent the ontology instance knowledge base, the tacit knowledge is obtained, and so as to serve as retrieval object. In this method the recall ratio of semantic retrieval is improved. Secondly, by borrowing idea from the Semantic Web vision and employing Semantic Web and MAS technologies, an architecture of Ontology- and MAS-Based Mechanical Design Intelligent Document(OMBMDID), which can enable intelligent agents to locate needed document segments oriented by meaning, grasp the knowledge contained in documents and put it into use automatically, is proposed. Firstly, the research context and positioning of this paper are discussed. Then, aiming at knowledge contained in mechanical design documents, semantic annotation granularity and semantic annotation categories are analyzed. On the basis of above analyses, the architecture of OMBMDID is proposed. The application analysis about semantic annotation, semantic retrieve and automatic utilization of document knowledge are also given.With methods proposed above as unit techniques, the semantic technology based mechanical design integrated system framework, which is designed to support innovative design, adaptive design and detailed design, is proposed. The components of this framework and its implementation techniques are also introduced. In the end, with bucket wheel excavator as case, the case study about above methods are given, the feasibility of semantic web technology based mechanical design is proved.In the end, the main study results and innovation points is summed up. The future research directions are also discussed.更多还原