【作者】 丘宏俊；

【导师】 陶华；

【作者基本信息】 西北工业大学 ， 航空宇航制造工程， 2006， 博士

【摘要】 由于飞机装配过程的复杂性，以及工艺设计与环境、经验相关的特点，开发智能化的飞机装配CAPP是一种典型的复杂问题，存在知识总结难、问题求解难、知识应用难等尚未解决的难题。随着飞机数字化制造工程的实施，如何开发飞机装配工艺设计的智能化系统，以提高装配工艺设计的效率和自动化水平已成为飞机研制过程迫切需要解决的问题之一。针对这一问题，本文以飞机部件装配工艺设计为研究对象，提出了“以知识库为基础，以推理机为核心，实现具有一定智能的计算机辅助飞机装配工艺设计”的构想。基于此构想，深入分析了飞机装配工艺设计过程，综合应用了本体论、知识工程和基于实例推理技术，灵活应用了XML技术，率先对装配知识表示、推理、应用等关键问题进行了系统研究，开发了基于知识的飞机装配工艺设计原型系统，旨在提高飞机装配工艺设计的自动化、智能化水平，使之成为飞机数字化制造的支撑。论文的主要研究工作如下：1、飞机装配工艺设计领域的知识本体模型研究了本体构建的原则与方法，提出了一种A2P2本体构建方法。严格按照A2P2法，深入分析飞机装配工艺设计过程，收集了大量飞机装配工艺设计领域的数据，在领域专家的参与指导下，对数据进行分析，研究了影响飞机装配工艺设计的因素以及装配工艺文件内容的基本形式，形成了装配知识的初步框架，为装配知识模型的形成奠定了基础。对初始知识框架进行转换、重构的概念化处理，建立了它们之间的联系，形成了领域术语。以领域术语为基础，提出了飞机装配工艺知识的本体模型，并对本体模型的合理性进行了评价。飞机装配工艺知识本体模型填补了在该领域知识本体研究的空白，也为装配知识表示、推理和应用奠定了基础。2、基于XML的装配知识表示及装配知识库构建基于框架表示和XML，研究了装配知识本体的形式化描述方法，设计了装配知识本体模型的BNF和XML Schema描述文件。全面比较、分析了XML与RDB在数据处理方面各自的优缺点，得出了在构建飞机装配知识库时必须同时使用XML和RDB两种数据处理技术的结论，并选用了Oracle XML DB作为装配知识库的底层支撑数据库。研究了XML与RDB之间的数据存储、映射策略，提出了一种混合式的飞机装配知识存储策略，并设计了装配知识库的逻辑结构。研究了装配知识库的索引机制，提出了一种递进关系的装配结构特征路径索引策略，构建了装配知识库的索引。应用XML和RDB着重解决了装配知识描述、组织和管理等问题。3、装配知识相似匹配推理技术基于框架表示知识推理和最近邻法的思想，设计了装配知识的匹配策略及算法过程，运用Delphi法与AHP法相结合科学地计算出各装配知识相似元的权重。分析了装配结构特征的特点，针对用树型结构描述的装配结构特征，提出了一种考虑语义、基于结构约束的装配结构相似匹配算法。分析了装配知识准确度的特征，提出了两种准确度的相似匹配策略，并建立了相应的数学模型。构建了装配知识匹配辞典，提出了一种基于匹配辞典的装配知识特征相似匹配算法，该方法从语义层面计算以字符形式描述装配特征的相似度。结合装配知识表示、知识库的特点及k-最近邻法的思想，设计了装配知识的推理过程，确定了搜索限制策略、冲突消解和求解策略，使得装配知识相似匹配推理在算法效率与匹配精度上达到较好的平衡。4、装配知识应用与知识库完善技术基于人机混合智能，研究了装配知识的复用过程、适应修改方法及知识学习方法。提出了一种基于系统问题求解经验的索引策略，并通过评分和时间戳的方式构建了索引，可以提高知识库和问题求解的质量。5、基于BOM的系统集成技术研究了飞机制造工程中xBOM的形成过程、MBOM与装配知识特征之间的关系以及装配知识特征的获取方法，设计了从MBOM中获取装配结构特征的算法过程，着重解决了待求解装配问题形成及系统集成的问题。最后，设计并开发了一个基于知识的飞机装配工艺设计原型系统，简称为KBA2P2(Knowledge Based Aircraft Assembly Process Planning)，对本文的研究思路、方法及关键算法进行了验证。结合企业的实际情况，对原型系统的推广实施进行了探讨。更多还原

【Abstract】 Process planning, which is condition-and experience-concerned, is a typical hardwork. For the complexity of aircraft assembly, aircraft assembly process planning(A2P2 for short) is usually considered as an "art" to be undertaken by good qualityand high technique. Consequently, the rules are not acquired easily or represented well,the reasoning is complex and the knowledge adaptation is difficult in developing theexpert systems of A2P2. With the implement of advanced manufacturing technologyin aeronautical enterprise, how to develop the intelligent systems of A2P2 to help toplan aircraft assembly process efficiently has become one of pivotal problems to beresolved. Aiming at this problem, a mechanism of Knowledge Based A2P2 isproposed in this paper. In this mechanism, assembly process planning of sub-assemblyand component is chosen as the research object, Ontology, Knowledge Engineering,Case Based Reasoning (CBR) are adopted as key technologies and XML is adopted asenabling technology, A2P2 knowledge’s representation, similarity measurement andadaptation are studied, a prototype (KBA2P2) is developed. To meet the advancementof A2P2, the main research issues and achievements in this paper are as follows:Ontology for A2P2 Knowledge One methodology of constructing ontology isproposed. Under the direction of this methodology, the course of A2P2 is analysed,plenty of data of A2P2 have been collected. With the help of experts in A2P2, theanalysis of these data is done, and the elements affecting A2P2 are separated, thenative knowledge framework of A2P2, which can’t be represented well and must beformalized, is gotten. After the transformation and reassembly of native knowledgeframework, the technical terms of A2P2, which can be represented in computerlanguage, are formed, and their relationship is set up. Based on these terms, theontology for A2P2 knowledge is constructed. This ontology lays the foundation ofrepresentation, reasoning and adaptation of A2P2 knowledge.XML-based Representation of A2P2 Knowledge and Construction ofRepository Based on Framework and XML, the representation of A2P2 knowledgeis researched, the BNF and the XML Schema of A2P2 knowledge are given. Theadvantage and disadvantage between XML and RDB (Relational Data Base) iscompared, the conclusion of combining XML with RDB to construct repository isdrawn, and Oracle XML DB is selected as the foundation database of repository. Thestrategy of data storage and mapping between XML and RDB is studied, a hybridstorage strategy of A2P2 knowledge is proposed, and the logical storage structure ofrepository is designed. The mechanism of repository index is researched, a specialstrategy of assembly structure feature of A2P2 knowledge is proposed, and the indexof repository is compiled. With the application of XML and RDB, the key problem ofA2P2 knowledge’s representation, organization and management is solved. Similarity Measurement of A2P2 Knowledge Based on the reasoning strategy ofFramework and the Nearest Neighbor,~☉the strategy and algorithm of similaritymeasurement of A2P2 knowledge is proposed. With the use of Delphi and AHP, theweight of knowledge feature in similarity measurement is calculated. After theanalysis of assembly structure feature described as tree model, a similaritymeasurement algorithm, which base on structure-and semantic-constraint of assemblystructure feature, is proposed, it can measure the similarity of assembly structurefeature efficiently. After the analysis of assembly tolerance feature, two similaritymeasurement strategies are proposed, and the mathematic models concemed arepresented. A mechanism of thesaurus to measure similarity of symbol assemblyfeature is proposed, the configuration of thesaurus and the algorithm concerned arepresented, this methodology can measure the conceptional similarity of char data typeknowledge feature. The A2P2 knowledge’s reasoning, retrieving and conflict strategy,which correspond to knowledge representation and repository structure, are planned.A2P2 Knowledge Adaptation and A2P2 Knowledge Learning Based on thehuman-machine hybrid intelligence, the adaptation and learning of A2P2 knowledge isresearched. An index strategy based on knowledge retrieving and adaptation ispresented, it uses Result Mark and Time Stamp to raise measurement efficiency andmanage A2P2 knowledge.BOM-based System Integration The evolvement of BOM (Bill of Material) ofproduct life cycle is studied, the relationship between MBOM and A2P2 knowledgefeature is researched, the acquirement strategy of A2P2 knowledge feature is studied,the algorithm to acquire assembly feature from MBOM is given. It proves that BOM isthe bridge of system integration.Finally, a prototype system, KBA2P2 (Knowledge Based Aircraft Assembly ProcessPlanning), is designed and realized to validate the whole thoughts and the keyalgorithms proposed. The prospect of the implement of KBA2P2 is forecasted.更多还原