【作者】 宋欣；

【导师】 郭伟；

【作者基本信息】 天津大学 ， 机械制造及其自动化， 2009， 博士

【摘要】 随着知识经济时代的到来,知识已成为产品创新和创造价值的源泉。特别是以大型装备为代表的复杂产品在分布式知识资源获取和重用方面的需求更尤为突出。为此,本文开展了对复杂产品设计知识进行合理有效地挖掘和重用的研究。对本文相关研究领域:知识工程、知识管理、产品设计过程、知识建模技术和设计知识重用方法的国内外研究现状进行了全面综述,总结并提出了这些领域目前存在的问题和未来发展趋势,明确了论文重点研究的内容:1、针对复杂产品协同设计中对知识资源产生的更深层次需求,提出了以面向复杂产品设计过程关键阶段的知识挖掘、知识模型构建及重用为核心的基于知识模型驱动的复杂产品设计方法,从总体角度论述了该方法实施的基本思想、体系框架及其各个子系统的基本职能。2、针对复杂产品设计过程关键阶段隐性知识的特点,提出了相应的知识挖掘方法。在需求分析阶段,通过粗糙集、QFD和DSM的集成,充分融合了用户需求,提高了复杂产品设计任务规划的准确性和合理性;在概念设计阶段,提出了基于粗糙集理论和贝叶斯决策的概念设计知识挖掘方法,对其中的一些数据挖掘算法进行了适应性改进,进一步提升了该挖掘方法的质量和效果;在详细设计阶段,提出了基于数值仿真结果的知识发现与繁衍方法,利用从大量仿真数据得出的回归模型和设计规则来指导方案的调整,有助于提高调整的正确性和效率。3、以实现知识的重用、共享、和集成为主导思想,对基于本体的知识建模方法进行了研究,定义了顶层复杂产品知识模型的概念实体及其关系,并构建出复杂产品需求分析、领域知识和过程知识的模型框架。4、开展了基于实例推理的知识模型重用方法研究,重点对模型的检索机制、相似模型匹配算法和调整机制进行了研究。建立复杂产品的扩展产品知识模型数组,以遍历该数组的方式实现模型检索;采用基于差异度的相似性度量方法来匹配相似模型,提高了检索效率和质量。提出了基于敏感性分析的属性权重算法,有助于更加准确地检索出相似模型;采用基于回归分析和规则推理的知识模型调整机制,降低了模型调整的难度和复杂程度,而且能够更有针对性,更好地适应新设计的要求,从而提升了方案的设计质量。在以上研究基础上,本文以灯泡贯流式水电机组这一复杂产品为研究对象,对关键设计阶段的隐性知识进行了挖掘,实现了知识模型构建和重用在水电机组关键部件设计中的应用,充分证明了基于知识模型驱动的产品设计方法在复杂产品设计中的实用性和技术可行性,有效地提高了企业研发和创新能力,显著地降低了成本,缩短了设计周期。更多还原

【Abstract】 With the era of knowledge economy, knowledge has become the headspring of production innovation and creating values. Especially, the development of complex production which represented by large-scale equipment has urgent requirements for distributing knowledge acquisition and reuse. Therefore, the effective methods of mining and reuse the knowledge in the process of complex production design were researched in this paper.The research status at home and abroad has comprehensive reviewed for the correlative domains with this research, such as knowledge-based engineering, knowledge management, production design process, knowledge modeling and design knowledge reuse methods. The key research contents were determined through the present problems and future development trends were summarized.1. According to the more profound requirments for knowledge resource application in the complex production collaborative design, a design method based on knowledge model driven was proposed. The knowledge mining oriented the key phases of complex production design process, knowledge modeling with its reuse compose the foundation of this method. As viewed from general, the basic idea of the method implementation, the system framework, and the basic function of four subsystems were demonstrated.2. All kinds of latent knowledge used and produced in design process of complex production with its reuse value were full analyzed. According to the characters of that knowledge in every design phase, the pertinence knowledge mining method was proposed. In requirements analysis phase, through the integration of rough set theory, QFD and DSM, the accuracy and rationality of the production planning can be improved due to the method could make the planning and structure design embody the customers’ requirements enough. In concept design phase, a concept design knowledge mining method based on rough set theory and Bayes decision was introduced. The mining quality and effect of this method was advanced further through adaptability improvement for some data mining algorithm in original theory. In detail design phase, a knowledge discovery and multiplication method based on numerical simulation results was put forward in order to fit the regression models and mine the latent design rules which could apply to direct the scheme adaptation accurately and efficiently. 3. With the knowledge model reuse, share and integration as the principal idea, a model building method based on ontology was proposed. The concept entities with their relations of top complex production knowledge model were defined and the model framework of requirements analysis, domain knowledge and process knowledge were structured.4. The research of the CBR method applied to knowledge model reuse was unfolded, its retrieval mechanism, similar models match algorithm and adaptation mechanism were mainly researched. The extended production knowledge model array of complex production was established, and the model retrieval was realized through the way of traversal this array. In order to advance the retrieval efficiency and quality, a similar model match method based on attributes diversity factor was applied. An attribute weight algorithm based on sensitivity analysis was proposed also make for correct retrieval similar models. Otherwise, a knowledge model adaptation mechanism based on regression analysis and RBR was proposed in order to reduce the adaptation difficulty and complexity, improve the design quality.Finally, with the bulb-tubular hydraulic generator as an example, some latent knowledge in key design phase was mined, and the knowledge models reuse of its key parts were realized. It is successfully verification that the method of knowledge model driven design in complex production development own practicability and technical feasibility. This method can effectively enhance enterprise development and innovation ability, greatly reduce production cost, and shorten design cycle.更多还原