Abstract: In order to explore the multidisciplinary design optimization strategy of centrifugal compressor and improve the optimization efficiency，an optimization system was development based on Proper Orthogonal Decomposition. Firstly，initial samples were selected to format the snapshot matrix. By retaining only the most significant components after POD analysis，a reduced order model of design space was constructed，in which the POD coefficients were acted as the new shape design variables. The method significantly reduces the number of design variables，linked with the multidisciplinary design optimization framework，offers a computational efficiency strategy for centrifugal compressor design. To verify the feasibility of the system，a compressor is optimized to achieve maximum isentropic efficiency at maximum speed state and cruise state. Results show the advantages of the scheme for effectively reducing the number of design variables，thus results into a faster convergence to the optimum. The isentropic efficiency of design point at two states has been improved by 3.5% and 4%，respectively.

Key words: Proper orthogonal decomposition；Optimization strategy；Centrifugal compressor；Multidisciplinary design optimization；Design space；Reduced order model

摘要： 为探索离心压气机多学科设计优化策略，进一步提高优化效率，建立了基于本征正交分解(POD)技术的离心压气机优化系统。首先选择样本数据生成快照矩阵，进行本征正交分解后，提取了占支配性的少数几个基函数，并以POD系数作为新的设计变量，构建了降阶的设计空间，从而大幅度减少了设计变量的个数。然后与离心压气机多学科优化流程相结合，建立了一种高效率的优化策略。为验证系统的可行性，以极大化等熵效率为目标，针对最大转速状态和巡航状态两个计算工况对压气机进行优化。结果表明:该优化策略在有效减少设计变量个数方面具有优势，从而使优化问题能够快速的收敛，优化后两个工况下的设计点等熵效率分别提高了3.5%和4%。

关键词: 本征正交分解；优化策略；离心压气机；多学科设计优化；设计空间；降阶模型