【作者】 汪旭；

【导师】 田凌；

【作者基本信息】 清华大学 ， 机械工程， 2010， 硕士

【摘要】 作为产品设计流程中最重要的环节之一,概念设计一直缺乏科学的方法论指导和计算机工具支持。当前广泛使用的几何造型概念设计均采用原始纸笔草绘的形式进行,自动化、信息化程度低,不能适应现代机械设计快速、复杂、创新等要求。以此为背景,本论文在国家863项目“支持产品创新设计的多学科协同建模/仿真方法及实现技术”的支持下,研究了基于设计意图捕捉的概念建模方法和关键技术,在此基础上开发了概念设计支持原型系统,解决了概念建模中对设计意图的快速固化和准确识别、基于语义的智能推送等问题,为概念设计阶段提供完整的支持。本论文的主要工作如下:1.提出了一种实验研究概念草绘行为习惯的方法,通过研究草绘速度变化趋势与设计意图之间的定性关系和草绘速度大小与设计意图表达精度之间的定量关系,建立了速度特征通道上的设计意图外化模型,解决了设计意图单通道识别不准确的问题。2.基于草绘行为习惯实验研究结果,提出了二维概念草图多设计特征识别方法,研究了概念草图设计中多个设计特征捕捉、分析、识别和整合等关键技术,大大提高了概念建模中设计意图识别的准确率和精确性。3.定义了一种新的支持语义计算的概念模型描述语言,实现了自动的二维概念模型的语义注入;建立了机械概念设计中的常用语义库,开发了描述语言的解析器和比较器,实现了语义层次上的概念模型理解。4.在概念模型语义理解的基础上,提出了一种基于设计意图捕捉的智能知识推送方法,设计及实现了主题爬虫和智能推送器:前者爬行抓取本地数据库和互联网相关知识,后者根据当前设计的语义信息进行知识推送,从而实现了与设计意图关联的知识推送。5.基于以上关键技术,设计开发了基于设计意图捕捉的概念设计支持工具,并进行了应用验证。更多还原

【Abstract】 As one of the most critical parts of the product design process, conceptual design always, since its first appearance, lacks of scientific methodology and computer support tools. The widely adopted approach of geometric modeling conceptual design, mainly with Pen and Paper sketch, is not able to meet the requirements of modern mechanical design to be rapid, complex and innovative, due to its low degree of automation and of information integration. With the support of the“863”National High-tech Project“Research on theory and realization of multidisciplinary collaborative modeling/simulation supporting innovative product design”, some methods and key technologies on conceptual modeling based on design intent capture are studied in this paper. A conceptual design supporting system which provides comprehensive support to conceptual design phase, and then realize the fast curing of design intent and accurate recognition and semantic-based intelligent push, etc. is developed consequently.Key achievements of this paper are as follows:1. An approach of experimental research on the behavioral action of conceptual sketch is proposed. By studying the qualitative relationship between the trend of sketch speed change and expression of design intent and the quantitative relationship between sketch speed and the precision of expression of design intent, a design intent externalization model on the channel of the speed feature is established and the issues on capturing the design intent only by single channel (the geometry channel) are solved.2. Based on the outcomes of the sketch behavioral experiment, a 2D sketch recognition method based on multi-design features recognition is raised. Key technologies such as capturing, analyzing, identifying and integrating multiple design intents in the conceptual sketch design are in-depth studied. As a result, the precision rate and accuracy of the conceptual modeling have been improved dramatically. 3. A semantics sketch description language called SSDL is defined and utilized to realize the automatic semantic injection into the 2D conceptual model. A common semantics library of mechanical design is established, parser and comparer for descriptive language are developed, and the semantic understanding of conceptual modeling is realized.4. Based on design intent capture and semantic understanding, a new intelligent knowledge pushing method is proposed, and the topic web crawler as well as an intelligent pusher are designed and implemented. The topic web crawler captures related predefined knowledge and information from local databases and the internet, while the intelligent pusher realizes design intent related knowledge push by pushing that based on the semantics information of current design.5. A conceptual design supporting prototype system basing on the key technologies mentioned above is designed and implemented. The application validation of the system is completed and described by the end of this paper.更多还原