【作者】 赵勇；

【导师】 王振国；

【作者基本信息】 国防科学技术大学 ， 航空宇航科学与技术， 2006， 博士

【摘要】 卫星总体设计属于典型的多学科问题。以多学科设计优化(MultidisciplinaryDesign Optimization,MDO)方法为核心实现设计优化与过程集成,对于提高卫星总体设计水平,实现卫星研制“快、好、省”的目标具有重要意义。论文以探索MDO方法与卫星总体设计过程相结合为目的,在系统研究MDO理论的基础上,建立了以分解、协调、搜索策略和MDO优化过程为核心的MDO理论研究主线,并将其应用于月球探测卫星和InSAR(Interferometric Synthetic Aperture Radar)卫星编队两类典型卫星总体优化设计问题。在MDO理论研究方面:首先,研究了基于图论的函数关系矩阵(Functional Dependency Table,FDT)与设计结构矩阵(Design Structure Matrix,DSM)分解策略,分析了基于超图的FDT分解模型和基于图论的DSM分解模型。算例测试结果表明:合理的学科分解有助于MDO问题快速准确地求解。其次,提出了基于改进Kriging模型的响应面协调策略。该策略遵循变复杂度建模思想,结合了二次多项式模型和Kriging模型的优势。三个不同复杂度的算例测试结果表明:该策略可明显提高响应面的近似精度和计算效率,并改善基于响应面方法的MDO优化过程的收敛性能。然后,研究了基于微粒群算法的设计空间搜索策略,提出了改进的微粒群算法(Improved Particle Swarm Optimization,IPSO)以及集成Powell法、模式搜索法与IPSO的混合微粒群算法。典型全局优化函数测试结果表明:两种改进算法在全局收敛性能和计算效率方面均有明显优势。最后,提出了针对BLISS 2000(Bi-Level Integrated System Synthesis 2000)优化过程的改进形式——HBLISS(Hybrid BLISS 2000)。HBLISS集成了基于改进Kriging模型的响应面协调策略和基于混合微粒群算法的搜索策略,并利用HLA/RTI(High Level Architecture/Runtime Infrastructure)实现了并行。算例测试结果表明:HBLISS在学科自治性和收敛性能方面较有优势,其并行实现可显著缩短计算时间。在MDO应用方面:首先,探讨了基于MDO的卫星总体设计过程的建模问题,深入分析了卫星总体设计过程中的总体技术流程、总体方案流程及其模型体系,提出了模型树与方案树的概念,建立了卫星总体MDO的基本框架。其次,综合上述研究成果,研究了月球探测卫星的MDO问题。针对此类以继承性设计为主的卫星总体设计问题,以单位信息量的成本最小为优化目标建立了总体参数优化模型,经合理学科分解后采用HBLISS进行集成和求解。优化结果较好地验证了HBLISS的可行性与有效性,并给出了较优的总体设计方案。然后,研究了InSAR卫星编队的MDO问题。针对此类以创新性设计为主的卫星总体设计问题,以全球高程测量为背景,分析了编队构形、SAR天线及卫星平台总体参数间的耦合关系,以系统成本最小为目标构建总体参数优化模型,在合理学科分解的基础上利用并行HBLISS进行集成和求解。结果较好地体现了MDO方法的优越性,并为进一步的设计奠定了较好基础。最后,在以上应用实例的基础上,建立了面向卫星总体的多学科集成设计系统,用于概念设计和初步设计阶段的卫星总体多学科设计、分析与优化。针对数字化设计系统的发展需求,提出了基于本软件系统构建卫星数字化集成设计系统的构想。总之,论文研究初步形成了比较完整的MDO理论研究主线,并将其应用于月球探测卫星和InSAR卫星编队的总体优化设计,为探索MDO方法在卫星总体设计中的应用进行了有益的尝试,也为进一步的MDO理论与应用研究奠定了良好的基础。更多还原

【Abstract】 Satellite System Design is a typical multidisciplinary problem. Multidisciplinary Design Optimization (MDO) is an effective method to resolve the design optimization and process integration, which is significant to improve the level of Satellite System Design and fulfill the object of "faster, better and cheaper" in satellites development. This paper aims at exploring the combination of MDO with the satellite system design process. After investigating MDO theories systematically, a MDO theoretical framework including the decomposition strategy, the coordination strategy, the search strategy and the MDO procedure has been established and applied in the system design of Lunar Explorer and InSAR Satellite Formation.Some MDO theories are discussed as follows:Firstly, two decomposition strategies------Functional Dependency Table (FDT) andDesign Structure Matrix (DSM) based on the graph theory are studied. The FDT decomposition model based on hypergraph and DSM decomposition model based on graph are analyzed. The testing results show that a reasonable disciplinary decomposition is beneficial to resolve some MDO problems quickly and exactly.Secondly, a coordination strategy of Response Surface Methodology (RSM) is brought forward on the basis of an improved Kriging model (VKRG-RS). It carries out the variable complexity modeling method and takes full advantages of the quadratic polynomial RSM and Kriging model. The testing results in three examples with different complexity show that this strategy can improve the approximation precision and computation efficiency of response surfaces and can also enhance the convergence of MDO procedures using RSM.Thirdly, as a search strategy of the design space, Particle Swarm Optimization (PSO) is discussed. An improved PSO (IPSO) and a hybrid PSO (HPSO) are presented respectively. The latter algorithm is a combination of Powell, Pattern Search and IPSO. The results of four global optimization functions show that two kinds of the new PSO do better in the global convergence and computation efficiency.Finally, an improved MDO procedure for BLISS 2000 (Bi-Level Integrated System Synthesis 2000) is proposed and named HBLISS (Hybrid BLISS 2000). VKRG-RS and HPSO are both integrated into HBLISS. The parallel computation of HBLISS is also implemented under HLA/RTI. The results show that HBLISS has the superiority in the disciplinary autonomy and convergence, and its parallel implementation can decrease the computation time greatly.Some MDO applications are discussed as follows:Firstly, for the satellite system design process, the pertinent problems about modeling are investigated. The system design flow, scheme forming flow and their model architectures are all analyzed deeply, the concept of model trees and scheme trees are brought forward, and then a basic structure of the satellite system MDO is built, as well.Secondly, the MDO of Lunar Explorer is studied in accordance with the above investigations. This kind of Satellite System Design mainly belongs to the inheriting design. A parametric optimization model of Lunar Explorer is set up with the cost per information as its objective. After a reasonable disciplinary decomposition, the MDO problem is integrated and resolved by HBLISS. Then, the results confirm the feasibility and validity of HBLISS and offer a better system design.Thirdly, the MDO of InSAR Satellite Formation with the mission of global elevation measurement is discussed. This kind of Satellite System Design mainly belongs to the innovative design. The couplings between parameters of the formation configuration, SAR antenna and satellite bus are analyzed. Then, an optimization model with the system cost as its objective is established and decomposed. Its integration and resolution are both implemented using parallel HBLISS. The corresponding results testify the superiority of MDO and lay the foundation for the further design.Finally, a multidisciplinary integrated design system is built for Satellite System Design. It can be applied in the multidisciplinary design, analysis and optimization of the concept and preliminary design phase. Then, on the basis of it, a preliminary architecture is conceived for the development of the enterprise-oriented digital integrated design system.To sum up, a relatively complete framework of MDO theories is formulated and used to resolve the system optimization design problems of Lunar Explorer and InSAR Satellite Formation. All of these are beneficial tries to explore the application of MDO in Satellite System Design and good foundations to further research on the theories and applications of MDO.更多还原