节点文献

基于知识融合的人机协同创新概念设计理论与系统研究

Research on Theory and System of Human-machine Collaborative Creative Conceptual Design Based on Knowledge Fusion

【作者】 赖朝安

【导师】 郑时雄;

【作者基本信息】 华南理工大学 , 机械制造及其自动化, 2003, 博士

【摘要】 目前, 传统的产品优化设计方法或单纯的人工智能优化方法已难以满足现代制造业对快速、创新设计的需求,因此研究新型的支持产品快速创新开发的现代设计方法并开发相应系统十分必要。本论文“基于知识融合的人机协同创新概念设计理论与系统研究”,既是对产品设计中知识工程的研究, 也是对面向工程广义优化的智能设计方法的研究, 对解决新产品开发决策支持问题具有重要的学术价值和实践意义。本课题得到了广东省科技创新百项工程资助项目(99B01901G)及广东省自然科学基金资助项目(990545)的资助。本论文对基于知识的人机协同创新方案设计理论与系统进行了系统的研究, 本文站在知识供应的高度考察了常用的智能优化设计方法, 指出产品快速创新设计的关键在于知识的供应特别是知识的融合与创新,并将设计知识适当分类, 并研究不同类型或同种类型知识之间的融合与创新的方法与途径。论文首先从总体上研究设计过程各阶段的决策内容及知识需求,对产品设计知识模型进行分析与综合研究, 对支持人机协同创新设计及知识供应的DSC 作开发上及理论上的规划。接着, 基于对知识的分类及知识融合与创新的目标展开如下具体工作: (1)提出并开发面向机器推理的XML 语义标记并用于消除知识文档中自然语言的语义歧义, 研究相应的推理技术及机械领域本体,开发基于自然语言WEB 的知识库系统, 以解决当前规则知识库知识共享与维护困难,知识库表现形式不统一等不足; (2)基于领域知识供应新方法,引入多Agent、CORBA 等技术开发网络化组件化接口标准化的领域知识供应Agent,实现领域知识基于网络共享的框架体系; (3)指出CBR 的缺点在于不能消除实例间的孤立状态,研究利用GA 的组合优化能力实现实例的融合与修改,实现实例知识的融合与创新; (4)研究能模拟发散性联想思维的、基于功能相似的、产品部件相似性计算方法,使它作为CBR 与GA 的结合纽带,并使CBR 与相似性计算为GA 生成具有高适应度并保持多样性的初始群体; (5)研究面向对象的柔性建模方法, 使人类专家可以实时地将领域

【Abstract】 At present, the classical optimization method and simple AI approach can’t satisfy the demands of rapid creative design of modern manufacture industry. Therefore, it is very necessary to research on new modern design method which supports rapid creative development of product, and also necessary to create corresponding software system. This paper “research on theory and system of human-machine collaborative creative conceptual design based on knowledge fusion”is to research on knowledge engineering in product design, also to research on intelligent design method for engineering generalized optimization. It has important academic value and practical worth for decision support in new product development. The study of this paper is supported by Natural Science Foundation of Guangdong Province (990545) and Science Innovation One Hundred Projects of Guangdong Province (99B01901G). This paper researches systematically the theory and system human-machine collaborative creative conceptual design based on knowledge, and studies some common intelligent optimization design methods at the level of knowledge supply, and points out that the key to rapid creative design lies in the supply of knowledge, especially in the fusion and creativity of knowledge. This paper also sorts the design knowledge and research the method and approach of fusion and creativity among knowledge of same or different type. First, as a whole, this paper researches decision content and knowledge demand in all design phases, and studies to sort necessary knowledge, and layouts DSC for human-machine collaborative creative design and knowledge supply in development and theory. Next, the following researches for goal of knowledge fusion and creativity based on knowledge sorting: (1) Research and develop XML semantic tags for machine reasoning, and use them to markup knowledge document in natural language, and research corresponding reasoning technology and machine domain ontology, and develop knowledge base system based on natural language, XML and WEB international standard, and solves those problems in knowledge share and maintenance and the drawback of non-uniform presentation scheme of classical knowledge base; (2) Based on new method of doma in knowledge supply, use multi-agent and CORBA technologies to develop agents for domain knowledge supply based on network, modules and standard interface, and realize the system frame for sharing domain knowledge based on network; (3) Points out that the drawback of CBR is due to the isolated cases, and studies to use the GA’s combination optimization ability to realize the fusion and adjust of cases, and then realize the fusion and creativity of cases knowledge; (4) Researches cluster analysis of product components based on simulating divergent human association thinking and similar function, and make it to be the linking between CBR and GA, and make CBR and cluster analysis produce original population with high fitness and variety for GA; (5) Research flexible modeling method based on OOT and huma n-machine incorporate, and let human expert turn his domain knowledge and design propose into optimization constrain model at real time, and realize the fusion and creativity among human intelligence, machine intelligence and many kinds of knowledge. At last, this paper researches the basic issues of creative design system, and puts forward a new method for automatic evaluation of creative conceptual design scheme, and puts forward and realize the integrate interface between conceptual design system and CAD system, and takes the injection mold design as example, and create integrated injection mold structure and injection parameter intelligent design system and puts it into use. The system can cuts down injection mold design cycle obviously, and aids designer in making decisions of new product development, and promotes design creativity.

节点文献中: 

本文链接的文献网络图示:

本文的引文网络