【作者】 曾喻江；

【导师】 胡修林；

【作者基本信息】 华中科技大学 ， 通信与信息系统， 2007， 博士

【摘要】 在过去的几十年里,地球观测、导航和通信得到了快速发展,人们希望能够通过各种手段在任意地点提供这些服务。利用卫星平台是实现这一目标的理想选择。几乎在相同的时间,利用卫星构建全球性地球观测、导航和无线通信网络等基础设施的需求增长,导致了卫星星座理论与技术的快速发展。卫星星座可以定义为按照一定规则构成、协同完成特定功能的卫星集合,是多颗卫星进行协同工作的基本形式。传统的卫星星座设计主要关心如何用尽可能少的卫星实现全球或纬度带覆盖。由于对称性,一般使用均匀分布的圆形轨道。这也便于得到比较容易处理的分析模型。但是,对于区域覆盖或混合轨道类型,该方法往往不再适用。不仅如此,现在的卫星星座设计很少再把单一的地球表面覆盖作为设计目标,更多的是考虑多个参数指标。这是一个多目标优化的问题。遗传算法(genetic algorithm,GA)已经被用于解决复杂问题。多目标遗传算法(multiple-objective GA,MOGA)适于解决有大量变量、需要进行多目标优化设计的问题。GA已经被用于非连续覆盖的的卫星星座设计。论文在分析卫星星座设计面临的技术挑战和现有工作不足的基础上,重点研究了基于GA和MOGA的卫星星座设计问题。论文的主要贡献是:在充分研究已有星座模型和GA特点的基础上,建立了通用6N星座模型(N为星座中的卫星数量)。该模型将每颗卫星的6个轨道参数都作为星座模型参数,去除了已有星座模型对卫星轨道参数的限制和约束,最大限度扩充了星座模型的参数空间,包含了所有已知的卫星星座构型以及其他可能的卫星轨道参数选择。传统卫星星座设计在优化设计之前需要选择合适的卫星星座构型。对于全球或纬度带卫星通信,选择哪一种卫星星座构型仍无定论。对于区域卫星星座或基于混合轨道类型的卫星星座,已知卫星星座构型往往无法得到最优结果。对于地球观测及各类科学实验卫星星座,已知星座构型更是难以胜任。基于通用6N星座模型,卫星星座设计者能够摈弃选择星座构型的步骤,直接根据卫星星座的设计需求在所有可能的轨道参数中进行优化设计,避免了星座构型选择对星座设计的负面影响。因此,6N星座模型具有最大程度的通用性,适用于绝大部分卫星星座设计任务。在通用6N星座模型的基础上,论文给出了改进的基于GA的无构型约束通用卫星星座设计方法,并首次提出了将代表MOGA领先水平的快速非支配排序遗传算法(the fast nondominated sorting genetic algorithm ,NSGA-II))应用于地球观测、卫星导航、卫星通信等不同卫星星座设计的统一算法模型,为复杂的卫星星座设计提供了强有力的多目标优化设计手段。6N星座模型在具备通用性优点的同时,不可避免的带来了优化参数空间过大的问题。由于范艾伦辐射带等因素的影响,卫星轨道参数还具有不连续、不可微等特点。GA和MOGA十分胜任这种优化问题。为了满足各类不同的卫星星座设计需求,论文在统一的卫星星座算法模型中给出了相应的基本性能评价准则。这些性能评价准则不仅仅局限于地理覆盖性能,还可以是用户提出的任意和卫星轨道参数相关的系统需求。设计者在进行自己的卫星星座设计任务时,只需要根据任务需求对卫星星座性能评价准则进行适当的修改或添加,就可以得到符合自身需要的(一组)最佳地球观测、卫星导航、卫星通信星座设计结果。通用6N星座模型和基于GA/MOGA的无构型约束通用卫星星座设计统一算法模型共同构成了完整的、适应现代卫星星座设计需要的、能够直接应用的卫星星座设计方法。论文研究了地球观测、卫星导航、卫星通信等卫星星座的系统要求,分别给出相对应的基本系统性能评价方法。为星座的优化设计提供了重要的优化依据,保证了设计结果的可行性。在卫星导航、卫星侦察、卫星移动通信方面等方面,论文分别进行了星座设计,给出了比已经公开发表结果更好的卫星星座设计结果。更多还原

【Abstract】 The development and growth of the earth observation, navigation and commu-nication during the past tens years has led to demand for and development of theseservices everywhere and over every possible medium. This includes the medium ofsatellite. During those same decades, the growth in use of satellite to provide a widely-available earth observation, navigation and wireless communications infrastructurehas led to the development of satellite applications using satellite constellation .A satellite constellation can be defined as a number of similar satellites, of asimilar function, designed for a certain purpose. Satellite constellation is a basalform for a group cooprating satellites.Traditional satellite constellation design has focused on optimizing global or zonalcoverage with a minimum number of satellites. In fact, in this case, the analysis ofthe configuration, based on circular orbits and uniform distribution of orbital planes,allows, because of the symmetry, to develop manageable analytical models. Notwith-standing, in some specific issues (i.e. regional coverage or hybrid orbit typies), eccen-tric orbits should be preferred for the coverage of restricted areas.Furtheremore, satellite constellation designers are rarely concerned with optimiz-ing performance with respect to a single objective such as earth coverage. Rather,multiple competing requirements drive the design.The genetic algorithm (GA) has been introduced as a robust technique to solvemany multivariable problems. The multiple-objective GA (MOGA) is used to solvethese problems because the multiple-objective optimal design is suited for problemswith large number of variables. MOGA for Walker satellite constellation and for zonalcoverage area is also introduced.Based on the analysis of the technology challenges for satellite constellation de-sign and the limitation of related work, this dissertation focuses on satellite constella-tion design optimized by GA and MOGA. The major contributions of the dissertation include:Considering the capabilities of GA and MOGA, the dissertation proposed ageneral 6N constellation model based on the analysis of some known constellationmodels. The general 6N constellation model parameters consist of all 6 orbitelements of every satellite. So the 6N model includes all known and unknownconstellation types. Constellation designers should select a constellation typefirst using traditional method. But the constellation designers could not seemto agree on the best approach to select the’best’constellation type for the sametask. The general 6N constellation model is foreign to constellation type, andsuits most satellite constellation design applications.The dissertation presented a improved satellite constellation design method basedon GA. Considering the earth observation, satellite navigation and satellite com-munication constellation design tasks, an one-up MOGA called the nondomi-nated sorting genetic algorithm 2 (NSGA-II) was used for a uniform constellationdesign algorithm model. Besides earth coverage, the uniform constellation de-sign algorithm model includes some basal system performance evaluation rulesfor earth observation, satellite navigation and satellite communication, whichcan be modified or added by constellation designer considering their own taskrequest.The dissertation analysed the characters of earth observation, satellite naviga-tion and satellite communication, proposed reasonable performance evaluationapproaches for vary satellite constellation systems.Some optimum satellite constellations, which better than the same task constel-lations in some publication, were demonstrated the the advantages of satelliteconstellation design method based on GA and MOGA for earth observation,satellite navigation and satellite communication.更多还原