航空学报 > 2010, Vol. 31 Issue (5): 940-945

基于混合人工鱼群算法的结构有限元模型修正

张安平1,2; 陈国平1   

  1. 1.南京航空航天大学 航空宇航学院 2.南京航空航天大学 无人机研究院
  • 收稿日期:2009-05-08 修回日期:2009-06-15 出版日期:2010-05-25 发布日期:2010-05-25
  • 通讯作者: 张安平

Structural Finite Element Model Updating Based on Hybrid Artificial Fish Swarm Algorithm

Zhang Anping1,2; Chen Guoping1   

  1. 1.College of Aerospace Engineering, Nanjing University ofAeronautics and Astronautics2.Research Institute of Unmanned Aircraft, Nanjing University ofAeronautics and Astronautics
  • Received:2009-05-08 Revised:2009-06-15 Online:2010-05-25 Published:2010-05-25
  • Contact: Zhang Anping

摘要: 将模拟退火算法(SAA)与具有交叉和高斯变异的人工鱼群算法(AFSA)相结合,提出了一种基于混合人工鱼群算法(HAFSA)的结构有限元模型修正方法;针对外编有限元模型修正程序直接嵌入Patran/Nastran软件存在困难的情况,设计了一种灵巧且方便的接口模块。以试验模型测试数据与有限元模型计算值的向量残差建立目标函数,在基本AFSA中引入交叉和高斯变异算子用于加快全局优化搜索速度,将目标函数优化值不断刷新公告板,再利用SAA进行局部细化搜索从而显著提高优化解的精度,在满足算法终止条件后获得设计参数的最优值;结合Fortran语言和Visual Basic语言编译接口模块,运行模型修正程序时循环修改Patran软件生成的建模文件并反复调用Nastran软件进行求解。以欧洲航空研究科技组织的基准模型——GARTEUR飞机模型为例,修正结果表明,应用HAFSA进行结构有限元模型修正是可行且有效的。

关键词: 模型修正, 接口模块, 优化, 人工鱼群算法, 遗传算法, 模拟退火

Abstract: By combining the artificial fish swarm algorithm (AFSA) with crossover and Gauss mutation with the simulated annealing algorithm (SAA), a novel structural finite element model updating method based on the hybrid artificial fish swarm algorithm (HAFSA) is presented, and a facile and convenient interface module is designed to deal with the difficulty that an external finite element model updating program encounters when it is directly implanted to the Patran/Nastran software. An objective function is established by using the resi-duals between the measurement data vectors of the test model and the calculation value vectors of the finite element model, and crossover and Gauss mutation operators are added to the original AFSA to increase the global optimization search velocity. The bulletin is refreshed by the optimization objective function value continuously, and SAA is applied to carry out local refined search to greatly improve the precision of the optimization solution. The optimization values of design variables are obtained after the algorithm end condition is satisfied. Fortran language is combined with Visual Basic language to compile the interface module. The Patran/Nastran software is transferred iteratively when the model updating program is run. A GARTEUR aircraft model is performed as an example, and updating results show that the finite element model updating based on HAFSA is feasible and effective.

Key words: model updating, interface module, optimization, artificial fish swarm algorithm, genetic algorithm, simulated annealing

中图分类号: