基于行为数据的设计知识提取及表达方法研究

项目名称： 基于行为数据的设计知识提取及表达方法研究

项目编号： No.61502285

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 计算机科学学科

项目作者： 崔嘉

作者单位： 山东师范大学

项目金额： 21万元

中文摘要： 设计知识在设计过程中起主导作用，但却因其不可触摸性难于在计算机中对各种设计知识进行演化、推理及预测应用。当前CAD研究主要通过对显性设计知识（几何、拓扑）进行计算和研究，但这种方法只能针对某种类数据进行计算，在适用性和灵活性方面有局限。这与设计知识本身的自相关、自演化的特点不相符。.本课题基于数据驱动研究方法，将设计行为引入知识体系中，通过邻接域相关性计算对设计知识在原有空间、拓扑关系的基础上，提出新的时序关系。通过DSR形状文法对设计行为以语言库的形式进行表示和建模。提出ACM规则应用模式和COC坐标体系能够方便的将设计行为映射到2D/3D空间的设计实体上；在‘User-interfered’衍生模式下，设计人员可以将新的设计知识通过设计行为与DSR知识库进行交互。在用户监督下，能够自适应的应用设计行为知识指导衍生设计。

中文关键词： 衍生设计；形状文法；设计知识表达；数据驱动；设计行为

英文摘要： Design knowledge is vital in design process. However, it is hard to be represented and used in design computation as the feature of intangible. In current CAD studies, the tangible design knowledge, such as geometry and topology, is widely used to compute design features. The computational constrains limited in specific data area reduce the applicable and flexible functions in real design works. As a result, the tangible design computation is conflicted with the self-correlation, self-improvement on the nature of design knowledge..In this study, in order to make some improvements on the problem mentioned above, the data-driven approach is used to represent design knowledge by design behaviors. Through the computing of neighbor relation from space, topology, the new time-order relationship is proposed. The DSR shape grammars represent design behaviors in terms of design rules and design languages. By using the ACM rule applying mode and the COC system, design behaviors can be mapped into 2D and 3D design entities. The ‘User-interfered’ generative design system release designers from heavy interactions. New design knowledge can be involved in design process between designer end and DSR shape grammars end. The new generations can be created under the guidelines of design knowledge model automatically or semi-automatically.

英文关键词： generative design;shape grammars;design knowledge representation;data driven;design behavior