【作者】 范培蕾；

【导师】 杨涛； 张晓今；

【作者基本信息】 国防科学技术大学 ， 航空宇航科学与技术， 2009， 博士

【摘要】 随着高超声速推进理论的不断深入,用于验证高超飞行器空中工作性能的高超声速飞行试验日益成为世界各国大力发展的重点方向。而设计综合性能最优的高超声速试飞器系统是保证试验可靠性与成功率的关键因素,因此,系统分析与设计过程中的多目标优化问题得到了诸多研究人员的普遍关注。本文以高超声速试飞器系统为应用对象,在综合考虑多项性能指标的同时,系统开展了多目标优化方法及其应用研究,取得了相应的研究成果。(1)在综述多目标进化机制的基础上,分析比较了常用的MOEA算法,构造分析了不同类型的测试函数,研究了算法性能评价方法。(2)研究了MOPSO算法。混沌变异机制引入到PSO算法中,克服了进化过程中出现的早熟收敛现象,改进了PSO算法的全局寻优能力;并将混沌变异机制应用于MOPSO算法,结合无变异机制的MOPSO算法,提出了分组MOPSO算法,解决了优化计算易陷入局部最优区域的问题;针对分布性与收敛性相互冲突而难以达到最优的难题,采用角度坐标方法与辅助适应度策略,提出了IMOPSO算法,更适合于飞行器系统等复杂优化问题的求解计算。(3)对Pareto最优解进行了深入分析。首先,建立了不同设计准则下的偏好函数模型,根据优化目标的取值范围确定了偏好区间;接着,依据优化目标的灵敏度分析,提出了Pareto最优解改进计算方法,给系统设计人员提供了更多满足偏好要求的候选解;最后,将目标总损失量作为方案稳健性指标融入决策过程,提出了基于目标总损失量择优的多目标决策方法,具有较强的工程应用性。(4)对高超声速试飞器系统进行了多目标优化设计研究。首先,概述了高超声速试飞器系统及其功能,确定了技术指标及结构组成;然后,对系统进行了多目标优化分析,研究了进行多目标优化设计的方法与思路;最后,开展了多目标优化方法在高超声速试飞器系统中的应用研究,包括:在详细分析动力模型、空气动力学模型、质量与结构模型、弹道计算模型等学科设计模型的基础上,以起飞质量、高超声速动力飞行段射程为目标函数,进行了多目标优化设计与决策分析,验证了进行多目标优化设计的必要性和合理性。(5)结合工程实际情况,研究了不确定因素影响下高超声速试飞器系统的多目标优化设计问题。在不确定多目标优化理论的基础上,综合考虑各学科设计模型中的诸多不确定性因素,建立了试飞器系统不确定多目标优化模型,采用基于概率支配关系的UC-IMOPSO算法进行进化计算,获得了稳健可靠的最终设计方案;并针对无控飞行方式下由不确定因素引起的弹道参数偏差较大的问题,提出了高度修正算法,提高了高超飞行器在正常工作动压范围内飞行的概率。本文工作是多目标优化方法在航天领域的一个典型应用,为其他具有不同试验任务的飞行器系统的优化设计提供了分析方法与研究思路。更多还原

【Abstract】 With the development of hypersonic air-breathing propulsion theory, hypersonic flight test is becoming an important direction for validating all kinds of hypersonic vehicles’performance indexes in complex flight atmosphere. Therefore, the design of hypersonic test vehicle with optimal performance indexes is commonly the key factor to maintain test reliability and success rate, so the multi-objective optimization problem in the process of system analysis and design has also been drawn attention by many researchers in aerospace design domain. This dissertation, focusing on the optimal design of hypersonic test vehicle based on various conflicting objectives, explores multi-objective optimization methods and their applications systematically and comprehensively. The results attainted are as followed.(1) On the basis of multi-objective evolutionary schemes and strategies, common MOEA algorithms and some kinds of test functions are expounded in detail, as well as their performance assessment methods of multi-objective optimizers.(2) Multi-Objective Particle Swarm Optimization algorithm (MOPSO) is studied thoroughly. Aiming to solve out the pre-mature convergence phenomenon, the chaos mutation is introduced into PSO for improving global optimal ability. Then, taking advantage of MOPSO algorithm with chaos mutation and no mutation, a kind of grouped MOPSO algorithm is proposed to tackle the problem of converging to local optimal region in the process of evolutionary calculation. For lessening their conflicts between diversity and convergence, a new improved MOPSO algorithm based on angle coordinate parameters and auxiliary fitness value is studied to maintain convergence indexes in the premise of good diversity for Pareto solutions.(3) Pareto Solutions and relative boundary are analyzed. Firstly, the preference function model is established with various design rules. Then, an improved method based on Pareto sensitivity analysis is proposed to increase the number of candidate set. Finally, the multi-objective decision making method based on multi-object-value loss is proposed with robust design analysis by the index of objective-value-loss, validating its rightness and effectiveness by the example of satellite appendix control system.(4) Some researches on hypersonic test vehicle’s multi-objective optimization design are carried out. Firstly, the overall design plan of low-cost hypersonic test flight is established, including the system function, indexes and configuration components. Secondly, the multi-objective optimization problem of hypersonic test vehicle is analyzed including the necessity and solving methods for multi-objective evolution. Lastly, the minimum takeoff mass and the maximum range of hypersonic free-flight period is considered as objective functions to carry out multi-objective evolutionary calculation on the basis of multidisciplinary knowledge such as propulsion model, aerodynamic calculation, ballistic calculation, structure and mass model. And the multi-objective optimization design and decision analysis to hypersonic test vehicle is carried out deeply, improving its necessity and reasonability for multi-objective optimization design.(5) In the light of engineering practical problem, the uncertain multi-objective optimization design for hypersonic test vehicle is investigated in this dissertation. Based on uncertain multi-objective optimization theory, the multi-objective optimization model of hypersonic test vehicle in the presence of uncertainty is established when considering all kinds of uncertain factors in multidisciplinary knowledge. And the robust ultimate design plan is ascertained by means of UC-IMOPSO algorithm with probability dominating relationship. Moreover, for the purpose of larger ballistic parameter deviations deriving from unguided flight motion, a new altitude-correction algorithm is proposed, which has increased hypersonic vehicle flight probability in the range of nominal working dynamic pressure.The research of this dissertation is a typically application of multi-objective optimization problem, which could provide some relative analysis and demonstrating methods for other aerospace vehicle systems with different flight mission.更多还原