面向设计流程前端基于统一可计算模型的创新产品集成优化设计方法

项目名称： 面向设计流程前端基于统一可计算模型的创新产品集成优化设计方法

项目编号： No.51275274

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 机械、仪表工业

项目作者： 张屹

作者单位： 三峡大学

项目金额： 80万元

中文摘要： 以概念设计为核心的设计流程前端是产品设计质量的决定性阶段。针对由于创新产品设计信息资源匮乏、现有设计模型信息表达局限，从而在概念设计中难以保证产品设计质量的问题，提出一种面向产品设计流程前端基于统一可计算模型的创新产品集成优化设计方法。通过研究基于多色集合(PS)的集成模型分层建模理论、方法，以及模型中各种复杂约束关系和参数特性的描述方式，建立涵盖设计流程前端的创新产品设计统一可计算模型；研究元胞多目标遗传算法(CMOGA)的模式理论、适应度评价、演化规则，开发具有稳定性、快速性的动态CMOGA；将统一可计算模型与CMOGA算法相结合，构建一种具有相当适应性、稳定性和快速性的混合优化算法框架，通过递阶结构PS模型缩小搜索的解空间，利用CMOGA算法保证种群的多样性和Pareto解集的构造，实现模型的高效优化求解；开发创新产品集成优化设计系统并在水下自主航行器的隔振系统设计中进行应用和验证。

中文关键词： 可计算模型；设计流程前端；创新产品；多色集合；元胞多目标遗传算法

英文摘要： The front part of design flow based on conceptual design is a crucial phase for the quality of product design. However, when the innovative products are designed, the quality can hardly be ensured during the conceptual design phase due to information resources starvation and information expression limitation of the existing design models. To solve this problem, a method of realizing global integrated optimization of innovative products based on unified computable model including the front part of design flow is proposed in this project. Firstly, a unified computable model of innovative products design including the front part of design flow is established through the analysis of Hierarchical modeling theories and methods of integrated model based on polychromatic sets (PS), and the description methods of various kinds of complicated constraint relations and parameter characteristic in the model. Then dynamic cellular multi-objective genetic algorithms (CMOGAs) with good stability and rapidity are developed after studying on the pattern theory, fitness evaluation and evolution rules of CMOGAs. Thirdly, a framework of hybrid optimization algorithm with considerable adaptability, stability and rapidity is constructed by combining the unified computable model with CMOGA, which is able to solve the complicated model

英文关键词： Computable model；Front part of design flow；Innovative products；Polychromatic sets；Cellular muti-objective GA