基于Kriging模型的离心叶轮结构优化设计

研究了以减重为目标的航空发动机离心叶轮结构优化问题,将Kriging模型与遗传算法相结合,应用拉丁方试验设计和有限元分析生成初始样本数据,利用初始样本数据建立离心叶轮重量和最大应力等状态参数的Kriging模型,运用遗传算法对该Kriging模型在设计空间进行全局寻优,利用有限元分析方法计算近似最优设计点的状态参数,并以此更新已有的设计样本数据,不断提高Kriging模型的近似精度.计算结果表明,基于Kriging模型-遗传算法的离心叶轮结构优化设计方法不仅可以获得良好的优化结果,比直接用遗传算法寻优减少了大量计算时间,提高了设计效率,同多项式模型-遗传算法相比也有效率优势.

Structural optimization of a centrifugal impeller based on a Kriging model

Based on a Kriging model and genetic algorithm, the structural optimization of a centrifugal impeller is studied in order to reduce the weight of an airplane. Obtained from Latin square experimental design and finite element analysis, the initial data is used to establish a Kriging parameters model of the weight and the maximum stress of the centrifugal impeller, which is then global optimized by genetic algorithm. The state parameters of the approximate optimum point are calculated with finite element analysis in order to up- date the existing sample data and to improve the approximate accuracy of the model. The results show that compared with direct searching with genetic algorithm, the combination of a Kriging model and genetic algo- rithm can get a good optimization result and reduce the computing time significantly. This method also has an efficiency advantage over the polynomial model and genetic algorithm.