基于模型重用与知识融合的复杂结构件可制造性优化方法研究

As simulation helps product designs satisfy functional requirements, manufacturability optimization make product designs ease to manufacture. Compared with simulation method like finite element analysis, there is a large gap between research of manufacturability optimization and requirements of improving reliability and shortening lifecycle of product design in an integrated manufacturing environment. This project takes typical complex structural parts in aerospace industry as an example, to propose new method based on model reuse and knowledge fusion, and to study basic theory of computer aided manufacturability optimization. It will explore content based retrieval theory of integrated information model, to reuse design data for filtering defect features of analyzed models. Relationships between reusable model features and semantic knowledge will be investigated, to visually integrate case model with multiple dimensional analysis rules and optimization solutions. By solving case driven rules of manufacturability analysis, semantic logic correlated model optimization approach will be explored. Based on previous research, this project transforms manufacturability analysis to multiple scale reuse of integrated information model, which would close the gap between model data and experiential knowledge, to improve the efficiency of manufacturability optimization of complex structural parts. The result of this project would be applied broadly in manufacturing industry such as aerospace company, to enrich theory of digital intelligence design, which would promote interdisciplinary innovation of knowledge engineering and design theory, and is of both important practical application and academic value.

正如仿真有助于产品设计满足功能要求，可制造性优化可使产品设计易于制造。与有限元等仿真方法相比，可制造性优化方法的研究同集成制造环境下提升设计可靠性、缩短开发周期的应用需求存在较大间隙。本项目以模型定义体系下典型航空复杂结构件为例，提出模型重用与知识融合的新方法，研究计算机辅助可制造性优化的应用基础理论；阐明集成信息模型内容相似性评估原理，重用设计数据聚焦模型缺陷特征；揭示可重用模型特征同语义知识的映射关系，在案例模型中可视化集成多维度的分析规则与优化方案；研究案例驱动下可制造性分析规则的求解算法，探索语义逻辑关联模型的优化途径。基于前期研究，项目将可制造性优化转化为集成信息模型的多粒度重用，有望弥补模型数据同经验知识的间隙，提高对复杂结构的可制造性优化效率。项目成果有望在航空制造等企业应用推广，丰富数字化智能设计理论，促进知识工程与设计领域的学科交叉创新，具有重要的现实意义和学术价值。

复杂装备离不开大协作，产品的设计与制造是其中尤其重要的阶段。但由于设计与制造分离，通过人工进行可制造性分析检查，存在效率和一致性问题。有效重用设计模型，显性化工艺知识，有助于实现工艺审查的自动化，以满足复杂产品设计制造协同的迫切需求。项目主要研究MBD体系下计算机辅助可制造性优化的应用基础理论。项目提出了基于模型重用与知识融合的新方法，将可制造性优化问题转化为集成信息模型的多粒度重用。通过主流工业软件二次开发实现了关键算法、知识库与推理机，构建了面向典型复杂结构件可制造性优化的原型系统。项目揭示了基于层次结构的相似性匹配的可制造性分析要素的解析机制；首次提出了基于模型局部特征的对称性定义，并在此基础上建立了交叉特征痕迹的识别算法，避免了采用图匹配的特征识别方法复杂度问题，最优复杂度降低为θ（N）；此外，建立了基于标准规范的加工能力的交互式可制造数据库维护方法，支持几何与非几何信息。项目提高了复杂结构件的可制造性优化效率，丰富了智能设计理论，促进知识工程与数字化设计制造的学科交叉创新，并为可装配性优化提供基础，具有重要的工程应用与学术价值。