【作者】 谢清；

【导师】 谭建荣；

【作者基本信息】 浙江大学 ， 机械设计及理论， 2007， 博士

【摘要】 针对产品设计过程中设计信息抽象性、模糊性和层次性等特点，提出了基于知识的定制产品功构映射方法与理论，并对定制产品功构模型构建、基于FCM的定制产品功构单元粒度划分、基于赋权自动机的定制产品功构映射过程以及基于BPNN&CBR的定制产品功构映射快速实现进行了深入研究，并结合实际应用项目将上述理论应用到具体产品设计过程中，取得了良好的效果，也证明了作者提出方法的先进性与正确性。论文的主要结构如下：第一章讨论了产品概念设计的特点及研究现状，研究了基于知识的定制产品设计理论，分析了现有面定制产品功构映射实现方法不足，论述了定制产品功能结构域映射的主要思想。第二章分析了定制产品的特点，研究了影响定制产品功构映射复杂度的主要因素，探讨了基于知识的定制产品需求表达方法，在此基础上提出了基于事物特性表的需求信息元模型；给出了基于事物特性表原理的需求信息元形式化表达方法及实现技术，从而构建了基于知识的定制产品功构模型(PFSM)；在公理化设计理论的基础上探讨了定制产品功构模型的构建原则及优化策略，及基于PFSM的定制产品功构映射实现流程。第三章基于构建定制产品功构模型，通过对模糊C均值算法中随机初始化策略进行改进，研究了功构单元粒度划分方法，结合定制产品设计过程中需求映射及实例映射特点，根据公理化设计原理探讨了基于信息熵的聚类评价函数构造。结合项目实施给出该方法在电梯制造业中的典型应用实例。第四章针对现有定制产品功构映射实现方法的局限性，给出了定制产品功构模型求解知识表达；根据模型层次节点功构单元间结构特征，提出了基于半环赋权自动机的单约束线结构产品功构模型功能结构域映射实现方法；按照“析取-合取”策略，运用自动机析取、同效、去冗余性等方法将多约束树型产品功构模型析取为若干相对独立的线结构产品功构模型，按单约束线结构产品功构模型映射方法求解，将所有解运用“合取”方法，进而实现多约束树型产品功构模型的功构映射。通过上述方法在液压机本体及移动工作台的功能结构映射具体应用实例，表明了方法的可行性与先进性。第五章根据定制产品功构模型的树结构特征，提出了基于结构图的定制产品功能结构树相似性算法，其通过最大同构子图获取算法、同构子图相似性算法实现产品功能结构树相似性比较，在此基础上，对功能结构树节点采用BPNN&CBR的相似性获取，即通过BPNN获取节点需求权重，并针对节点特征值类型特点，给出不同类型特征输入与样本实例特征差分距离的计算公式，按照实例推理方法获取与用户需求匹配的节点实例。最后给出了YF32-3150 KN液压机本体功能结构树相似性计算案例，及基于实例的移动工作台特征权重获取的具体应用。第六章依托国家科研项目，结合企业具体应用，开发了定制产品功构映射平台，将产品功构模型的映射理论和方法应用于产品需求获取、产品模块粒度划分、产品设计任务信息管理、产品设计模块信息管理、产品设计流程管理以及产品基础数据管理等过程中，给出了系统平台的应用背景、实施策略、体系结构和功能模块。第七章总结了本文的主要研究内容和成果，并给出了今后的研究方向更多还原

【Abstract】 A new method and theory, which is about the mapping between function domain and structure domain of customization products, is presented in this dissertation , mainly consisting of the customization product function-structure model, the granularity of the product function-structure units base on the algorithm of fuzzy cluster means, the mapping procedure between function domain & structure domain based on the weight automata, the implement of mapping between function domain & structure domain based on artificial neural network and case-based reason. Furthermore, the effective applications on several research projects proved advancement and validity as well.The frame of this dissertation is as follows:Chapter 1 discusses characteristics and status quo of the product concept design, makes research on the theory of customization product design based on knowledge, analyzes the deficiencies in existing mapping methods between function domain & structure domain for customization product. Furthermore, the main ideas are given on the customization product’s mapping between function domain & structure domain.Chapter 2 analyses the characteristics of customization products, makes research on the main factors which influence the complexity of the customization product’s mapping between function domain and structure domain, discusses the representation of customization product requirement, and product requirement information meta model based on the tabular layouts of article characteristics is also presented, then the method and technology of requirement information meta’s representation and implementation is given based on the tabular layouts of article characteristics, which leads to the construction of the customization product function-structure model. The construction principles and optimization strategies of product function-structure model is discussed based on the axiomatic theory. Besides, the flowchart of product configuration design base on the product function and structure model is given.Chapter 3 makes research on function-structure unit granularity by improving the random initialization of Fuzzy Cluster Means which is based on the customization product function and structure model, moreover, function of granularity has also been constructed as associated with domain mapping according to the axiomatic theory and the characteristics of requirement mapping and instance mapping during the process of mapping. Moreover, a typical instance is given related with elevator manufacturing project.Chapter 4 proposes the knowledge representation of customization product function-structure model, aiming at the limitation to implement the mapping of customization product from function domain to structure one. With the structure features of the model above, the mapping implementation for the single-constraint product function-structure model is represented based on semi-ring weight finite state automata. According to the disjunction-conjunction strategy, the multi-constraint product function-structure model can be extracted to several independent single-constraint ones with methods as automata disjunction, equivalency and off-redundancy etc. After all single-constraint ones are solved respectively, all solutions to them are conjunctive which attain the mapping implementation from function domain to structure domain for multi-constraint product function-structure model. In addition, a typical concrete instance is given related with the project of hydraulic machine, which shows the validity and promise of the method mentioned above.Chapter 5 puts forward a similarity algorithm of customization product function-structure tree based on graphics, according to the tree-structure characteristic of customization product function & structure model, which implements the similarity of product customization template by the largest subgraph isomorphism attainment and the similarity algorithm of the subgraph isomorphism. Next, the similarity access of nodes in the template is realized as bellow: attain the requirement weight of nodes through artificial neural network; construct the formula of the distance between the feature input and the instance feature with the different characteristic value type. Then, the instance corresponding with the requirement is gotten by the method of CBR. Besides, a typical instance is given related the body of hydraulic machine which type is YF32-3150KN and the moving workstation of hydraulic machine.In chapter 6, the mapping platform between customization product function & structure is developed and applied in some project, which shows the validity and feasibility of the new theory and method of product design knowledge in the dissertation.In chapter 7, conclusions are given along with recommendations for future research.更多还原