【作者】 侯文君；

【导师】 廖启征； 孙汉旭；

【作者基本信息】 北京邮电大学 ， 信号与信息处理， 2010， 博士

【摘要】 机械产品的复杂性和装配工艺规划的经验性,导致了虚拟装配的高度复杂性。针对“虚拟装配信息空间和工艺信息管理任务日益增长的复杂度”与“交互通道的带宽和人有限的认知以及信息处理能力“之间的的矛盾,本文将普适计算“以人为中心”的思想引入虚拟装配环境,提出了“以情境为基础,以感知为核心”实现“具有一定智能的计算机辅助装配工艺规划系统”的构想。以系统实现为目标,综合应用本体论、人工智能、Agent技术为手段,并灵活应用XML这种技术,从情境表示、情境构建、情境推理、情境应用及系统集成五个依次递进的层面,对基于情境感知的虚拟装配工艺规划进行系统的探索和研究。通过3D-IVAPP原型系统的开发,对论文提出的理论技术和关键算法进行了集成验证。论文主要工作如下：1、基于本体的情境感知体系结构框架构建在分析了装配工艺规划的影响因素以及装配工艺知识分类基础上,讨论了装配情境的内涵与外延,得出了装配情境与装配知识的关系,结合CTK构建了基于本体的情境感知体系结构框架,提出了面向虚拟环境智能的小世界情境的概念,并基于本体设计了装配情境工作模型和感知框架,为智能装配系统制定了功能模块与关键技术。2、智能装配的情境信息的本体建模通过对传统CAD系统装配设计方法的分析,设计了一种虚拟装配的多层次装配模型,并在此基础上重点分析了装配特征语义信息的提取,提出了一种可操作性强的层次化装配语义模型；在对本体的概念、构造准则方法以及本体语义分析的基础上,建立了装配的本体领域词汇；建立了装配特征与本体的信息映射关系,基于SolidWork进行了语义信息提取,并应用集合理论封装了情境装配信息模型,为系统的情境感知奠定了基础。3、本体语义推理与智能扣件全面分析、比较了OWL-DL和SWRL在语义推理方面各自的优缺点,提出了基于OWL-DL和SWRL结合的DLRL推理方案,进行OWL本体知识库和SWRL规则库的建立,设计了事例与规则的转换方法,并建立了零件装配体从属关系、概念一致性检测、装配体自由度等规则描述；应用DLRL推理器,提出了本体实例化智能扣件Widgets,给出了智能零件识别的算法流程为挖掘隐性知识提供了有效手段。4、装配情境感知机制研究基于虚拟装配的功能特点,指出任务感知是虚拟装配情境感知的核心,分析了装配中的任务、角色之间的关系,构建了面向装配情境感知的多Agent系统结构,给出了基于多Agent的的消息响应模式,设计了基于多Agent的装配情境感知机制。在分析情境识别、获取需求基础上,建立了装配情境多粒度模型,提出了基于商空间的多粒度感知算法。并利用粒子群计算对装配序列进行了智能规划,为智能虚拟装配系统的感知决策提供了方法,并提高了效率。最后,设计并开发了一个基于情境感知的虚拟装配工艺规划原型系统,简称为3D-IVAPP,其中包括：从SolidWorks环境中导出三维模型、语义信息,Open Inventor支持下的虚拟环境中智能装配扣件的算法流程和实现以及基于Web的工艺资源管理平台中用户数据、权限管理以及三维模型的共享和可视化。对本文的研究思路、方法及关键算法进行了验证。更多还原

【Abstract】 The complexity of machinery products and the empirical of the Assembly process planning Lead to a high degree of complexity of virtual assembly. For " the growing complexity of virtual assembly information space and process information administrative tasks" and "the contradiction between the interactive channel with limited bandwidth and human limited cognition and information processing capabilities.In this paper, "The thinking of "human-centered" in Pervasive Computing was inducted into the virtual assembly environment, putting forward the "context-based, perceived as the core" to achieve a certain intelligent computer-aided assembly process planning system concept. To achieve the system, it comprehensively applied Ontology, Artificial Intelligence, Agent technology as tools, and flexibly applied XML technology.From the context expression construction context, context reasoning, context applications to system integration five step-up level,it explored and researched the virtual assembly process planning system based on context perception systematicly. By 3D-IVAPP prototype system exploitation, the paper presents the theory and key algorithm integrated authentication. The paper’s main works are as follows:1. The Construction of Ontology-based Context Aware FrameworkBased on the analysis of the impact factors of assembly process planning and the classification of knowledge of the assembly process, the connotation and extension of the assembly context were discussed, the relationship between the knowledge of assembly and assembly context was presented, the assembly context hierarchical model was established, Ontology-based context aware framework was built combining with CTK, by proposing the concept of intelligent small-world situations faced to virtual environment, designing the framework and working model of Assembly context based on Ontology, constituting the function module and key technologies for the intelligent assembly system2. The Ontology Modelling of Intelligent Assembly context information Through the analysis of the traditional method of assembly design CAD system, the multi-level assembly model was designed based on virtual assembly, which the assembly semantic information extraction process was analyzed on the basis of; After analysing the ontology concepts, construction guidelines methods, and Ontology Language Semantic,the establishment of this assembly model in the ontology areas vocabulary given the concept of assembly-level ontology relations; it analysed the information map of the CAD system to the semantic model, established the process of the semantic information extraction algorithm based on solidwork by giving the example applications, and application of set theory, encapsulated the assembly context information model using the set theory.3. Ontology reasoning and intelligent WidgetsBy a comprehensive analysis, comparing the OWL-DL, and SWRL reasoning in the semantics of their strengths and weaknesses, it designed DLSL SWRL reasoning program based on OWL-DL combining to SWRL, established OWL ontology knowledge base and SWRL rule base, which were inducted into Jess inference engine, described the rules of parts-assembly belong-to relationship, concept of consistency testing the degrees of freedom of assembly. After the instance analysis by using the DLRL reasoning engine, the ontology instances of intelligent fastening method was proposed, it presented intelligent parts identification algorithms, benchmark parts and parts to be installed intelligent aware algorithms based on assembly tasks, which provided an effective means of perception of intelligent assembly tasks.4. Assembly context-aware mechanismsBased on the needs analysis of context recognition acquisition, this paper proposed a multi-granularity design of the context, and set up a multi-granularity module of the assembly context, and introduced a multi-granularity algorithm based on entropy space. And in this paper, based on the features of virtual assembly, the task-aware is pointed to be the core of the virtual assembly context aware, and the relationship between the tasks and the roles during assembly is analyzed, what’s more, multi-agent system and its request response model is build, and the assembly context aware mechanism for multi-agent system is proposed.Finally, based on the context aware, a virtual assembly process planning system model (3D-IVAPP) is designed and developed, including 3D virtual model from the solidworks, semantic information, smart assemble fasteners algorithm in virtual environment with open inventor, user data, rights management, sharing and visualization of 3D model on the Web platform of technology resource management. What’s more, the research ideas, methods and the key algorithm are introduced at the end.更多还原