空气动力学
离心式压缩机
涡轮机械
计算机科学
还原(数学)
气体压缩机
等熵过程
灵敏度(控制系统)
Sobol序列
计算流体力学
拉丁超立方体抽样
数学优化
参数统计
围带
轴流压缩机
参数化设计
航空航天
总压比
流量(数学)
熵(时间箭头)
优化设计
多目标优化
压力降
控制理论(社会学)
航天器
埃菲
数学
可制造性设计
最优化问题
工程设计过程
气动弹性
作者
Z. Shen,B Wang,F. Jiang,J. Wu,S. Miao,X. Du
标识
DOI:10.47176/jafm.19.9.4396
摘要
This study proposes a sensitivity-driven design-space reduction and surrogate-assisted optimization framework for sCO₂ centrifugal compressor impellers. A parametric aerodynamic database was constructed using Bezier curves and optimal Latin hypercube sampling, enabling an efficient exploration of the high-dimensional design domain. A hybrid feature selection strategy integrating a Sobol sensitivity analysis and random forest ranking was developed to identify dominant design variables, address the high dimensionality, and reduce the computational cost. Subsequently, a surrogate-assisted optimization algorithm was employed to maximize the pressure ratio and isentropic efficiency while maintaining robust aerodynamic performance. CFD validation confirms that the optimized design achieves improvements of 3.29% in the pressure ratio and 5.31% in isentropic efficiency under the design conditions, with consistent gains across the 0.7Q–1.3Q flow range. An analysis of the flow mechanism reveals that the improved performance stems from mitigated secondary flows and uniform blade loading, resulting in a 6.82% reduction in average entropy generation and overall aerodynamic losses, thereby demonstrating the strong potential of the proposed framework for advanced turbomachinery design applications.
科研通智能强力驱动
Strongly Powered by AbleSci AI