【作者】 张宝；

【导师】 滕弘飞；

【作者基本信息】 大连理工大学 ， 机械设计及理论， 2007， 博士

【摘要】 卫星舱布局设计是基于卫星公用平台的卫星总体设计的重要内容。通常，它是指在卫星舱内外如何对卫星的各种仪器、设备进行布置，以满足各种工程技术约束条件并尽可能对布局方案的各项性能指标进行优化。卫星舱的布局设计对于缩短卫星设计周期、节约成本、提高卫星的性能等方面有着重要作用。在数学上，它属于组合最优化问题；在工程上，属于复杂工程系统问题。面临的主要困难在于，既涉及数学上的组合爆炸问题，又涉及工程复杂性问题，并要达到工程实用。本文以委托项目“航天器布局优化设计与仿真系统研究与开发”为工程背景，在国家自然科学基金的资助下，研究高效、实用的卫星舱布局优化求解算法。主要工作包括以下两个方面：(1)提出一种基于子种群节点的金字塔模型粒子群算法(Pyramid model based particle swarm optimization，PPSO)，用于求解规模较小的卫星舱布局设计问题。粒子群算法(PSO)简单易用、收敛速度较快，但存在着易于早熟等不足，为此本文给出了两种改进方法：一是利用基于金字塔模型的多种群搜索来提高种群多样性；二是对丧失搜索能力的退化粒子进行变异操作，以扩大粒子的搜索空间。与粗粒度遗传算法相对应，可以认为本文PPSO属于一种粗粒度粒子群算法。通过对5个经典函数无约束优化算例、2个Packing算例的数值仿真实验结果表明了该算法的可行性。(2)提出一种带启发式规则的协同进化粒子群算法(Co-evolutionary particle swarm optimization with heuristic rules，CEPSO_HR)，用于求解规模较大且适合系统分解的卫星舱布局设计问题。为此首先将卫星舱布局问题分解为几个小规模的子布局问题，然后采用协同进化的框架，并将粒子群算法与布局启发式规则相结合进行求解。本文启发式规则主要包括待布物的定位跳跃、定向排斥和换位3种启发式规则，用来提高算法的计算精度、计算效率与计算鲁棒性。研究了CEPSO_HR的种群分解方式、合作个体的选择、适应度函数的协作评价等问题，尤其对如何利用启发式规则生成待布物的新位置进行了详细阐述。最后，通过简化返回式人造卫星回收舱、简化国际商业通信卫星舱布局设计的数值仿真实验，验证了本文上述两种算法的可行性和有效性。本文在理论上研究了一种新的粗粒度粒子群算法和一种带启发式规则的协同进化粒子群算法，有助于粒子群算法和协同进化算法的深入理论研究；在工程上，本文算法为“航天器布局优化设计与仿真系统”提供技术支持，并可望推广应用于其它类型的航天器布局相关设计。更多还原

【Abstract】 The layout design of satellite module takes an important role in the satellite schematicdesign based on the public platform. In general, this layout design concerns how to locate theapparatus and equipments in the limited space of a satellite module, satisfying variousbehavioral constraints of the interior and exterior environment and optimizing theperformance indexes of layout scheme. This study has the important affections for reducingthe design period, saving the cost, improving the performance of the satellite, and so on. Inmathematics, the layout design of satellite module belongs to a combinatorial optimizationand NP-hard problem. In engineering, it belongs to a complex engineering system. The maindifficulty for solving this problem is that it is related to not only the combinatorial explosionin mathematics, but also the engineering complexity. Moreover it needs to be applied topractical engineering.Taking the project, i.e. the study and development of design and simulation system forthe satellite layout optimization, as engineering background, and being supported by theNational Nature Science Foundation of China, this dissertation studies the fast, practicallayout optimization algorithms. The main contributions are as follows:(1) A pyramid model based particle swarm optimization (PPSO) is presented, where anode of pyramid model corresponds to a subpopulation. This algorithm is used to solve thesmall-scale satellite module layout problems. Particle Swarm Optimization (PSO) is easy touse and its convergence speed is fast, but PSO is easy to fall into premature, so two improvedmethods are given out here. First, multi-population search based on pyramid model is givenout to enrich the diversity of the population. Second, mutation operation is executed by theparticles that have no ability to evolve, and its aim is to expand the search space of theparticles. Reference to coarse-grain genetic algorithm, PPSO can be taken as a kind ofcoarse-grain particle swarm optimization algorithm. The experimental results showed thatPPSO is feasible in five classic functions and two packing problems.(2) A co-evolutionary particle swarm optimization with heuristic rules (CEPSO_HR) ispresented. This algorithm is used to solve the large-scale satellite module layout problemswhich are suitable to decompose. The satellite module layout is decomposed into severalsmall-scale layout problems. Then the co-evolutionary framework is adopted, and PSO andheuristic rules for layout are integrated to solve this problem. Here, heuristic rules mainlyinclude position jumping, direction repulsion and transposition rules of objects, and they areused to improve the calculati0nal precision, efficiency and robustness of the algorithm. It isstudied the decomposition of the population, selection of the collaborative individual,collaborative evaluation of the fitness of CEPSO_HR, especially explained how to generatenew positions of objects using layout heuristic rules.At last, through the numerical experiments of the recoverable satellite and internationalcommunication satellite, it is verified of the feasibility and validity of the two algorithmspresented by this dissertation,In theory, this dissertation studies a coarse-grain particle swarm optimization algorithmand a co-evolutionary particle swarm optimization with heuristic rules. It is helpful to the further study of theory for PSO and co-evolutionary algorithm. In practice, the algorithms canafford the technology support for "the design and simulation system for the satellite layoutoptimization", and it’s hoped that they can be applied to the layout design of other types ofspacecraft.更多还原