【作者】 邓亚权；

【导师】 薛彩军；

【作者基本信息】 南京航空航天大学 ， 飞行器设计， 2010， 硕士

【摘要】 结构疲劳失效对飞行安全危害极大,因此开展长寿命设计技术的研究具有重要现实意义。本文针对结构疲劳寿命优化中疲劳寿命估算耗时较长等缺点,提出一种基于代理模型的结构疲劳寿命优化方法,并开发了结构疲劳寿命优化专用软件,将其应用于飞机起落架结构优化中。提出基于代理模型的结构疲劳寿命优化理论框架。首先,通过对结构整体的疲劳寿命分析找出危险子部件,应用动力学仿真分析获得危险子部件的载荷谱。然后,建立危险子部件疲劳寿命和重量的代理模型,优化危险子部件的疲劳寿命。对三类常用代理模型和生成样本点的试验设计方法进行较为系统的比较分析,研究发现以拉丁超立方试验设计构建径向基函数模型具有较高的精度和拟合效率,较为适用于构建疲劳寿命的代理模型。提出最优信息概念,将其引入粒子群算法中,解决了基本粒子群算法后期由于粒子都向局部最小收敛的问题。应用Java语言开发了部件疲劳寿命优化的专用软件,实现了试验设计、径向基函数代理模型的构建和基于粒子群优化算法优化功能。以某型飞机前起落架为例,按照本文的结构疲劳寿命优化框架,对前起落架整体进行疲劳寿命分析,找出了危险子部件——前撑杆,用动力学仿真分析获取了前撑杆的载荷谱。通过PCL语言建立了前撑杆的参数化有限元模型,构建了前撑杆疲劳寿命和重量的代理模型,最终实现了疲劳寿命优化,其疲劳寿命从1.51×105提高到7.24×105。更多还原

【Abstract】 The fatigue failure may do great harm to flight safety, so it has important practical significance to carry out the research about improving fatigue life. Structural fatigue life optimization is practical methodology to deal with this issue, but the fatigue life analysis in optimization cycles is a time-consuming procedure. To overcome this disadcantage, a method of structural fatigue life optimization based on metamodels is proposed. Based on this method, a optimization software was developed, and it is applied to the structural optimization of a nose landing gear.A metamodel-based theoretical framework of structural fatigue life optimization was proposed. The fatigue life analysis of the whole structure was carried out to figure out the critical component with the lowest fatigue life, then the load-time history of this component was obtained by dynamics simulation analysis. After that, the metamodels on fatigue and weight of the critical component were constructed, and then its fatigue life was optimized.Three kinds of metamodels and two kinds of design of experiments was analized systematically. It is found that the radial basis functions constructed by Latin Hypercube design is the best metamodel for the fatigue life. This is because that it is not only a good model with a higher fitting efficiency, but also has a better adaptability.The“Optimal Information”was introduced into the partical swarm optimization to deal with the problem that convergence speed was slowed down significantly in the end due to all of the particals converge to the minimum. Then structural fatigue life optimization software was developed using Java, and this optimization software can carryout the design of experiments, construct the metamodels and provide the modified partical swarm optimization.Taking a nose gear as an example, the critical component:forward strut was figure out by fatigue life analysis of the nose gear, then the load-time history of forward strut was obtained by dynamics simulation. Based on parametric finite element model of forward strut established by PCL, the metamodel on fatigue life and weight of forward strut was constructed. Finally, the fatigue life of forward strut was optimized, and the fatigue life of the forward strut was improved significantly, from 1.51E5 cycles to 7.24E5 cycles.更多还原