【作者】 乔志宏；

【导师】 隋允康；

【作者基本信息】 北京工业大学 ， 固体力学， 2002， 硕士

【副题名】用PCL对MSC/NASTRAN进行优化模块的二次开发

【摘要】 本文利用MSC／PATRAN提供的PCL语言，针对桁架结构的截面优化对MSC／NASTRAN优化模块进行了二次开发。 首先，对于应力约束，采用满应力设计方法，以各杆的截面面积为设计变量，通过迭代求解，得到最优设计结果。其中，在每一步迭代过程建立优化模型之前，采用射线步技术，调整了结构的性态，保证了收敛的稳定性。 其次，对于桁架结构中受压杆件的失稳破坏，本文提出采用相似变换方法找到截面积与惯性矩的关系，根据压杆的临界应力分别导出大、中、小柔度的迭代公式，设计出压杆的截面积。 第三，对于应力约束、局部稳定约束和位移约束的问题，本文采用序列二次规划SQP方法进行了求解。把复杂的应力约束和局部稳定约束转化为动态尺寸约束，引入倒变量x_i＝1/A_i将目标函数展开为二阶近似，将位移约束用莫尔积分化为一阶近似，用对偶规划方法将原问题化为等价的二次规划问题，调用Lemke算法，求得最优设计结果。 第四，将建立优化模型的菜单融合到PATRAN界面，实现程序的可视化。用户可以通过人机界面的交互功能确定设计变量、约束条件及目标函数三要素，实现了优化模型建立的柔性。 本文利用PCL语言对上述优化模块在NASTRAN上进行了程序实现，多个算例验证了程序具有合理性、精确性和收敛速度快等特点。更多还原

【Abstract】 Based on MSC/NASTRAN software, this paper uses the PATRAN/PCL (PATRAN Command Language) to secondly develop a new software of optimum design module for sectional optimization of truss structures.First, considering the stress constrain, the full stress design method is used to solve the sectional optimization of truss structures, in which the section area is taken as design variable. The final design result can be found after iterations. Before the optimization modeling, the scaling step technique is used to adjust the structure behavior, and the stability of computational convergence is guaranteed.Second, to solve the local stability problem of the bar under pressure, a function relationship between the sectional area and the inertia moment is established by using the similar transformation. Accounting to the critical pressure stress, iterative formulas are derived which include big, middle and small flexible degree.Third, for the problem of the stress, local stability and displacement constraints, the sequential quadratic programming(SQP) method is adopted in this paper. The stress and local stability constraints are transformed intomovable lower bounds of sizes. An inverse variable xt = ?is inducted, andthe objective function is expanded as second order Taylor approximation while the displacement constriction is expanded as first order Taylor approximation. The Lemke algorithm is used to get the final design result.Forth, visualization is realized by setting the menu into PATRAN. The users can define design variable, constraints condition and objective function by the alternative function of computer interface, by which the optimization module can be set up expediently.In this paper, optimization modules mentioned above are all carried out in NASTRAN by using of PCL language. And several numerical examples are used to testify the program credibility and validity.更多还原