基于结构层自底向上聚类重构的应用机构创新设计理论与平台

Creative design of constituent mechanisms through bottom-up clustering and reconstructing of structural layer is a research area that has not been paid much attention to, internationally. However, it has critical theoretical and practical values for further improving the ability for the function synthesis method to obtain creative mechanism design schemes, and to improve the market competiveness of products. Based on the previous research and the quotient space synthesis theory, the three-dimensional bottom-up clustering synthetic model of behavior fragment collections in form layer is established. Through clustering of the set of behavior fragment collections which come from the multi-function behaviors of constituent mechanisms, a number of complete behavior groups are constructed to match the sub-functions in function layer and to be realized by the multi-function behaviors of the constituent mechanisms in form layer. And the coordination among behavior groups is guaranteed during the clustering process. In this way, the complex functional requirements can be met by a small amount of constituent mechanisms, and the new way to carry out the innovation of the constituent mechanisms through bottom-up clustering and reconstructing of the behavior fragment collection set is opened up. On this basis, a computational platform to support the creative design of practical mechanisms based on the bottom-up clustering and reconstruction of the behavior fragment collections in form layer will be constructed. Focusing on some typical engineering applications, the application of the platform to deal with practical engineering problems of mechanisms will be studied for improving the engineering application capacity and the ability of handling on practical engineering problems of the platform.

结构层自底向上聚类重构构成机构的创新，是国际学术界尚未涉足的研究领域，其对于进一步提升功能综合法获得创新性机构设计方案的能力，提高产品的市场竞争力，具有重要的理论与工程应用价值。本课题在前期研究基础上，基于商空间合成理论，建立结构层行为片段集合自底向上的聚类合成三维空间模型，使来源于结构层匹配机构多功能行为的行为片段集合聚类重构，构成既能与子功能匹配，又能被实用机构主、次要功能所实现的完整行为群体，且各行为群体有机协同，用少量构成机构满足复杂功能需求，开辟结构层自底向上聚类重构应用机构创新的新途径。在此基础上，建立支持自底向上的基于行为片段集合聚类重构的应用机构创新设计自动化平台，结合典型工程应用领域，开展应用机构创新研究，提高平台的工程应用适用性与处理实际工程问题的能力，为相关产业的产品创新设计提供工具支持。

借助于计算机强大的计算与存储能力，创成不低于人类专家创新设计水平的机械系统，是计算机辅助创新工具研究领域追求的终极目标；而解决缺少构成推理知识的并联与复合组合机构的功能综合，则是机械学领域的国际性难题。学术界广泛采用基于FDM方法构造计算机化工具，因其功能与解密切相关，导致所构造的工具创新空间探索能力不足，且极易产生解固化。本课题采用与解无关的细粒度功能单元（SKB）描述FBS模型中的子功能，基于商空间等价划分模拟人类创新思维过程（RBP方法），其中，基于粒度空间的划分用于模拟发散思维过程；基于商拓扑与商属性的正则化处理，用于模拟收敛思维过程，使得到的行为子集能够有机协同地实现总功能需求。在此基础上，采用自底向上SKB聚类重构与变异操作（R&M方法），进一步拓展功能层与结构层的创新解空间。研究表明，RBP方法具有探索最高新颖性-功能创新解空间的能力，R&M方法具备获得历史性功能创新（H-Novelty）解方案的能力。特别是，R&M方法能够对人类专家实施并联与复合机构巧妙组合进行模拟，创成用更简单构成结构实现复杂功能要求的机械系统，进而极大提升机械产品的市场竞争力。由此，本课题解决了基于FDM方法计算机化工具长期以来存在的解固化、创新解空间探索能力不足的难题。在计算机辅助创新领域取得了突破性重大进展，使利用计算机模拟人类创新思维，挖掘不低于人类专家创新水平的机械创新方案成为可能，将有力地促进智能CAD的发展。