基于结构代理模型与二级优化策略的航天结构模型修正研究

Updating finite element dynamic models of aerospace structures based on the surrogate model and 2-level optimization strategy

随着我国航天事业的不断发展,航天结构形式越来越复杂,有限元结构模型往往存在各种假设和简化,因此与实际结构往往存在一定的误差,特别是部件的连接、边界条件的不确定以及材料性能(尤其是复合材料)与工艺的不确定性,都会对有限元模型的分析精度产生较大影响。本文基于初始有限元模型,通过试验设计构造结构代理模型,然后采用遗传优化算法与梯度优化算法的二级优化策略进行模型修正。另一方面,在每次模型修正迭代分析之前,自动进行置信值MAC(Modal Assurance Criterion)分析,使有限元分析模型与实验结果进行匹配,提高模型修正的正确性。分析表明该修正方法具有较高的分析精度,也能对结构参数进行识别。

With the development of China＇s aerospace industry, aerospace structures are becoming more complex. Assumptions and simplifications are often used for the finite element model（FEM）,such as the connection stiffness of components, the effect of boundary conditions and uncertainty of material properties, therefore, there are some errors in the FEM. Based on the original FEM, the surrogate model can be established by design of experiment（DOE）, and the 2-level optimization strategy can be used for model updating, which includes genetic algorithm and Gradient algorithm. On the other hand,before each iteration of model updating, the MAC（Modal Assurance Criterion） should be analyzed automatically, so that there is a match between the method has a higher accuracy,and FEM model and result of this experiment. Analysis showed that the the structural parameters can also be identified by using the method.