面向产品族设计的主从关联优化问题与模型方法研究

发布时间：2018-05-26 06:32

本文选题：产品族 + 主从关联优化　；参考：《天津大学》2014年博士论文

【摘要】：相比以往单层模型或者分阶段的研究，产品族设计由于其多主体、多层次和多学科的特点，，其中的某些问题更适合利用主从模型进行设计，也越来越得到重视。本文从产品族设计的主从关联优化问题入手，首先对产品族主从关联优化模型进行了研究，然后在此基础上对主从优化模型提出了分解协调方法，最后对产品族设计中存在的两类典型主从关联问题进行了详细研究。全文每章节研究分别配以数学算例、通用电机产品族和自行车产品族进行了案例说明。论文的具体工作包括：第一、本文对产品主从优化模型从分析框图、数学模型及一般分类框架等方面进行了理论研究。其中分析框图建立了二维分析模型，数学模型中根据产品族本质属性，研究了具有协同项特征的主从优化模型，最后通过从者之间变量和策略集的划分，给出了模型的一般分类框架。第二、对工程中具有层次结构的设计优化问题，目标级联是一个有效的分解方法。本文针对下层为凸规划的一类双层规划问题，在利用KKT条件转化为单层优化问题后，基于目标级联方法并结合目标函数和约束条件的FDT表，提出了一种两层结构的目标级联式分解方法。第三、在前面产品族一主多从模型一般形式的基础上，针对从者之间决策变量出现耦合的情形，利用在目标函数上添加偏差项，给出了一个分解协调的算法，使得从者之间变量独立，并讨论了原问题的Pareto解与分解协调后问题可行解之间的关系。第四、针对产品族的模块配置为主，属性参数设计为从的一类典型设计问题，建立了一个关联和协同优化的混合0-1非线性双层规划模型：模型的上层是整个产品族的配置优化问题，以成本-效用为优化目标，决策产品族的模块属性配置；下层是产品模块属性的参数优化问题，以属性的技术性能为优化目标，决策与上层配置相匹配的属性参数设计。第五、针对模块属性组合中的一类典型设计问题，在模型中引入了0-1变量实现对模块属性档级的选择，以制造商成本和客户效用的比值及产品族通用性指数作为上层目标，以产品可选属性效用最大化为下层目标，实现必选模块和可选模块的主从优化组合。
[Abstract]:Compared with the previous single-layer model or multi-stage research, product family design is more and more suitable for master-slave model design because of its multi-agent, multi-level and multi-disciplinary characteristics. Starting with the master-slave association optimization problem of product family design, this paper first studies the master-slave association optimization model of product family, and then puts forward a decomposition coordination method for the master-slave optimization model. Finally, two kinds of typical master-slave correlation problems in product family design are studied in detail. Each chapter of this paper is illustrated by mathematical examples, general motor product family and bicycle product family. The specific work of the thesis includes: Firstly, this paper makes a theoretical study on the product master-slave optimization model from the analysis block diagram, mathematical model and general classification framework. In the mathematical model, according to the essential attributes of product family, the master-slave optimization model with the characteristics of cooperative items is studied. Finally, by dividing variables and policy sets between followers and followers, a two-dimensional analysis model is established. The general classification framework of the model is given. Secondly, objective cascade is an effective decomposition method for the design optimization problem with hierarchical structure in engineering. In this paper, for a class of bilevel programming problems with convex programming, after transforming the KKT condition into a single-layer optimization problem, a FDT table based on the method of cascading objectives and combining objective functions with constraints is presented. An object cascade decomposition method with two-layer structure is proposed. Thirdly, on the basis of the general form of the previous product family, a principal and multiple slave model, a decomposition and coordination algorithm is given by adding deviation items to the objective function in view of the coupling of decision variables between the follower and the follower. The relations between the Pareto solution of the original problem and the feasible solution of the problem after decomposition and coordination are discussed. Fourthly, for the module configuration of the product family, the attribute parameter is designed as a typical design problem. A hybrid 0-1 nonlinear bilevel programming model of association and cooperative optimization is established. The upper level of the model is the configuration optimization problem of the whole product family. The optimization goal is cost utility and the module attribute configuration of the product family is decided. The lower layer is the parameter optimization problem of the product module attribute. The attribute parameter design is based on the technical performance of the attribute and the matching between the decision and the upper configuration. Fifthly, aiming at a class of typical design problems in module attribute combination, 0-1 variable is introduced into the model to select the module attribute file level, which takes the ratio of manufacturer cost and customer utility and the generality index of product family as the upper target. With the aim of maximizing the utility of optional attributes, the combination of required module and optional module is realized.
【学位授予单位】：天津大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TB472

【参考文献】