【作者】 史彦军；

【导师】 滕弘飞；

【作者基本信息】 大连理工大学 ， 计算机应用， 2006， 博士

【摘要】 本文以我国一类航天器布局设计为工程应用背景,在国家863计划航天领域高技术项目的子课题“航天器布局优化设计平台研究与开发”和国家自然科学基金的资助下,研究复杂布局的协同差异演化(Differential Evolution)方法及其系统开发。 我国航天技术及其产业的蓬勃发展对航天器布局优化的理论方法研究及其工程实用化提出了更高的要求。航天器舱布局优化是研究在满足各种工程技术要求的条件下,如何对航天器舱内外的各种仪器和设备进行优化布局。航天器舱布局问题在数学上属组合最优化和NP问题,在工程上属复杂系统工程问题,既存在数学上的计算复杂性,又存在工程上的工程复杂性,这是一个至今尚未很好解决的问题。研究该问题的求解方法是一个具有挑战性的、开放性的重要课题。 1995年Storn和Price给出的差异演化是一种基于群体差异的计算智能类优化算法,具有全局并行快速搜索的特点,已成功应用于各种全局优化问题。协同演化(最早在1991年由Hillis提出)是一种在现有演化算法基础上形成的一种解空间分离编码的动态搜索和优化技术。本文在对一类航天器布局问题分析的基础上,结合上述两种技术,研究协同差异演化理论及其在该布局问题上的应用,主要有以下三个方面的工作: (1) 给出了一种协同差异演化方法(简称CCDE)。该方法在差异演化算法的基础上,研究了合作式协同演化的复杂问题求解框架,将复杂问题空间分解为多个子空间并分配给相应的子种群,然后对各子种群内部采用差异演化算法的交叉、变异和选择操作进行演化,对于各子种群之间则以协同方式共同演化。此外,探讨了不同的问题分解方式、各子种群的协作方式和个体适应度的计算方法。数值仿真实验结果表明,协同差异演化方法的计算性能优于协同遗传算法、差异演化算法和遗传算法,具有较好的应用前景。 (2) 给出了一种协同混合遗传差异演化方法(简称CGDE)。首先研究混合遗传差异演化算法(简称GDE),在GDE算法中差异演化和遗传算法共用一个种群。该算法首先采用差异演化算法的交叉、变异操作生成新个体,随后对该个体施加了遗传算法的高斯变异操作,达到保持种群多样性和待布物策略性振荡的目的(针对布局问题),再应用差异演化的选择操作选拔较优个体构成新种群,并以此循环演化。进而,针对诸如航天器的复杂工程布局问题的特点,基于协同演化方法和GDE算法,给出了协同混合遗传差异演化方法。对典型布局问题的仿真实验结果表明,GDE算法优于差异演化算法和遗传算法;若采用CGDE方法,则其优于协同差异演化方法和协同遗传算法。 (3) 给出了一类航天器舱布局优化模型、求解方法和软件系统。首先给出了本文航天器布局优化模型和求解方法(包括上述方法的工程应用),进而开发了航天器舱布局优化软件,重点讨论了其系统架构,以及软件开发的关键技术(包括开发模式的选择、用户交互界面的开发和自动装配定位技术的实现)。最后以返回式人造卫星回收舱和国更多还原

【Abstract】 At the engineering background of a class of spacecraft layout design, this dissertation aims to provide more efficient layout optimization for layout problems of complex spacecraft using Cooperative Co-Evolutionary Differential Evolution (DE for short). Additionally, National Nature Science Foundation of China and project of Research and Development of a Software Platform for Layout Design of Spacecrafts from China Aerospace Science and Technology Corporation supported this work.The rapid development of space technology and industry has need of more researches on the theories and methods of layout optimization for spacecraft. Spatial layout design is to study how to arrange the instruments and equipment inside (or outside) the vessel while satisfying all kinds of engineering constraints. This layout problem is known as a combinatorial optimization and NP problem in mathematics, and complex system problem in engineering. To solve this challenging and open problem is very important to space industry.DE is proposed by Storn and Price in 1995, which is a stochastic, population-based, and relatively unknown evolutionary algorithm for global optimization that has recently been successfully applied to many optimization problems. Moreover, co-evolution (proposed by Hillis firstly in 1991) is a dynamic optimization method based on current evolutionary algorithms, using split search space and gene coding. This dissertation focuses on how to apply DE and co-evolution to layout problems, including the following contributions:(1) A cooperative co-evolutionary differential evolution (called CCDE) was proposed. Based on DE, CCDE uses a problem-solving framework based on cooperative co-evolutionary approach, and partition a high-dimensional complex search space by splitting the solution vectors into smaller vectors, then use multiple cooperating subpopulations (or smaller vectors) to co-evolve subcomponents of a solution. Additionally, this dissertation also discusses the different methods of problem-decomposition, collaboration methods of each sub-species, and evaluation methods of individual fitness. Lots of numerical experimental results showed that CCDE has a marked improvement in performance over the DE, CCGA (cooperative co-evolutionary genetic algorithm) and GA (genetic algorithm).(2) A cooperative hybrid genetic differential evolution algorithm was proposed. Firstly, a hybrid genetic differential evolution algorithm (GDE for short) was proposed. In GDE algorithm, DE and GA share one common population. GDE generates new individual using mutation and crossover operators of DE. Subsequentially, this new individual is changed by Gauss mutation of genetic algorithms to keep the diversity of population and oscillation oflayout objects. Thirdly; a better individual is selected using selection operator of DE to form a new population. Finally; a loop using the above steps is taken to evolve. Moreover; according to the characteristics of layout problems; based on GDE and cooperative co-evolutionary algorithms; a cooperative GDE is also presented. The experimental result using canonical multi-container layout problem showed that GDE achieved better layout results compared to DE and GA; and cooperative GDE is better than CCDE and CCGA.(3) The optimization models; problem-solving methods (including engineering application of the abovementioned methods); and layout software for a class of spacecraft layout was proposed. Firstly; an optimization models and problem-solving methods for spacecraft layout problems were proposed. Then; optimization software for spacecraft layout was developed. Thirdly; optimization software architecture was proposed; and the key software development technologies were introduced; including development mode; user interface and automatic location. Finally; taking the simplified retrievable artificial satellite and international communication satellite INTELSAT-III as engineering example; the experiments showed that the proposed optimization methods and software were effective to the layout problem of this dissertation.This work is expected to help to improve the theory of DE and co-evolution and make the abovementioned methods be practical in engineering. Moreover; users from space industry had checked the proposed software prototype for spacecraft layout problems. Furthermore; this work would be extended to apply in layout of tank; mechanical engineering; robot; rocket; submarine and under-water suspension engineering; etc.更多还原