【作者】 刘克龙；

【导师】 姚卫星；

【作者基本信息】 南京航空航天大学 ， 飞行器设计， 2007， 博士

【摘要】 多学科设计优化(MDO)是飞行器设计的一个重要发展方向。自从MDO被提出以来,得到了国内外学者的重视,并获得了较大的发展,但仍然存在某些问题,如:多学科设计优化方法过于复杂;代理模型技术仍然以多项式响应面为主,难以描述复杂的耦合关系;代理模型建立过程中对异常点的处理考虑较少,可能会影响代理模型的精度;利用MDO对机翼进行多学科设计优化时结构和气动两个子系统还难以并行,需要建立机翼上准确的气动载荷分布函数等。本文对多学科设计优化技术的几个研究方面,如MDO方法、代理模型技术、气动和结构学科分析优化模型的建立等问题进行了研究。在协同优化方法(CO)的基础上,通过减少设计变量个数,调整子系统目标函数等,提出了低自由度协同优化方法(LDFCO),利用三个数学算例对其计算效率进行了检验,结果表明低自由度协同优化方法可靠性较高,其计算量比CO方法明显减少。以异常点检验技术和序列Kriging优化(SKO)为基础,根据MDO中代理模型的特点,提出了一种面向MDO的代理模型建立方法-基于SKO的代理模型。研究了子系统的不同层次代理模型的建立方法,并成功地应用到了机翼多学科设计优化过程中。提出了分步SKO方法,并以该方法为基础,提出了一种基于分步SKO方法、遗传算法、MSC. Patran和MSC. Nastran的结构布局优化方法。利用该方法对复合材料加筋板和复合材料机翼进行了优化设计,达到了较好的优化效果。建立了基于Catia、Gambit和Fluent等通用软件的气动自动化分析模型。提出了利用三系数四次响应面模型拟合机翼上的升力分布以实现不同子系统并行和气动力的传递。以气动自动化分析模型和三系数四次响应面气动载荷分布模型为基础,提出了面向高精度气动分析模型的气动分析代理模型的建立方法。利用低自由度协同优化方法对轻型飞机机翼进行了气动/结构多学科设计优化,气动和结构两个子系统实现了并行,其中气动分析分别采用涡格法程序VLM和CFD软件Fluent,结构分析优化采用有限元软件MSC. Patran&Nastran。结果表明低自由度协同优化可以成功地应用于机翼多学科设计优化中。更多还原

【Abstract】 Multidisciplinary design optimization (MDO) is one of the important development directions of aircraft design. Since MDO has been proposed, it is getting more attention and becomes prevailing. However there are still some problems. Multidisciplinary design optimization methods that appeared in recent years are too complicated. Polynomial response surface approximations are still the main surrogate models applied, but it is difficult to describe complicated relations with response surfaces. How to deal with outliers gets little attention when surrogate modes are constructed. It is difficult to make structure subsystem and aerodynamics subsystem concurrent when a wing is designed with MDO without accurate aerodynamic load distribution functions.Several research fields of MDO such as MDO methods, surrogate models, analysis and optimization models of structure and aerodynamics were investigated in this thesis.Low Degree-of-freedom Collaborative Optimization (LDFCO) was proposed through reducing the number of design variables and adjusting the objectives of subsystems based on Collaborative Optimization. Three test examples were used to verify the efficiency of LDFCO. The results show that LDFCO has high reliability and needs much less computation cost than CO. Based on outlier test and Sequential Kriging Optimization and according to the characteristics of surrogate models in MDO, a construction method of surrogate model for MDO–SKO based surrogate model was proposed. Several surrogate models of different levels of subsystems were researched and were applied to multidisciplinary design optimization of wings successfully.Two-step SKO was proposed and a structural layout optimization method was built based on two-step SKO, genetic algorithm, MSC. Patran and Nastran. A composite stiffened panel and a composite material wing were optimized with the structural layout optimization method. The results show that this optimization method has the ability to find out the approximate optimal solution, and can be used in engineering design conveniently.Based on Catia, Gambit and Fluent, an aerodynamics automatic analysis model was proposed. In order to transfer aerodynamics load among different subsystems, a quartic response surface with three coefficients was developed to fit lift distribution. Based on the aerodynamics automatic analysis model and the quartic response surface with three coefficients, a construction method of aerodynamics analysis surrogate models for high fidelity aerodynamics analysis models was proposed.Low Degree-of-freedom Collaborative Optimization was applied to solve Aerodynamic/structural multidisciplinary design optimization of a wing. The structure subsystem and the aerodynamics subsystem are parallel. Aerodynamics performance was calculated with VLM or Fluent. Structure analysis was performed using MSC. Patran and Nastran. The results show that Low Degree-of-freedom Collaborative Optimization can be applied to multidisciplinary design optimization of wings successfully.更多还原