基于自适应本征正交分解的涡轮级多学科设计优化

Multidisciplinary Design Optimization for Turbine Stage Based on Self-Adaptive Proper Orthogonal Decomposition

为了提高涡轮级多学科设计优化的优化效率,基于本征正交分解(Proper Orthogonal Decomposition,POD)技术,并结合快照样本自适应更新方法,提出了一种综合的涡轮级多学科优化系统。首先,通过进行POD分析,仅保留占优势的基函数,并以POD系数作为新的设计变量,设计变量个数由60个缩减为5个,提高了优化效率。然后,基于自适应进化规则,优化过程中对快照样本进行不断的进化和修正,从而提高POD精度。最后将该方法与涡轮多学科优化流程相结合,建立了一种高效率、高精度的优化策略。某涡轮优化的结果表明:该优化策略适于设计变量较多的复杂优化问题,且具有良好的收敛性,优化后设计点等熵效率提高了3.5%。

In order to improve the optimization efficiency of the multidisciplinary design optimization of turbine stage,a hybrid optimization system was development based on Proper Orthogonal Decomposition（POD）combined with self-adaptive snapshot updating technique.Firstly,by retaining only the most significant components after POD analysis,the POD coefficients were acted as the new design variables.The number of design variables has been reduced from 60 to 5.So the optimization efficiency was improved.Secondly,the snapshot was evolved and modified iteratively using self-adaptive evolutionary algorithm during the optimization process.The precision of POD analysis was improved.When linked with the multidisciplinary design optimization framework,a computationally efficiency and high precision strategy was offered for turbine stage design.A turbine optimized results show that the strategy is suitable to complex problem with large number of design variables,and it has good convergence.The isentropic efficiency of design point has been improved by 3.5% after optimization.