基于遗传算法和逐次序列二次规划的叶栅基迭优化  

BLADE STACKING OPTIMIZATION BASED ON GA AND SQP METHODS

作　　者：赖宇阳[1] 袁新[1] 

机构地区：[1]清华大学热能工程系,北京100084

出　　处：《工程热物理学报》2003年第1期52-54,共3页Journal of Engineering Thermophysics

基　　金：国家重点基础研究专项经费资助项目(No.G1999022306);国家自然科学基金资助项目(No.50076019)

摘　　要：本文提出并实现了一种叶轮机械全三维优化设计体系。该组合优化方案基于遗传算法(GA)进行全局搜索,响应面(RSM)近似建模和逐次序列规划(SQP)局部寻优,采用高阶非均匀有理B样条进行三维叶片参数化造型和高精度、高分辨率Navier-Stokes求解程序进行粘性流场数值模拟。通过对一低展弦比大折转角的透平环形叶栅进行基迭优化,结果表明该组合优化方案具有很好的效率、可靠性和全局搜索能力。A hybrid optimization approach for automatic design of 3D turbine blades was proposed. The Genetic Algorithm (GA), the quadratic Response Surface Model (RSM) and the Sequential Quadratic Programming (SQP) were used for global exploration, approximation and local refinement respectively. The three-dimensional viscid flow field within turbomachinery was simulated by a parallelized CFD code for solving compressible, three-dimensional Navier-Stokes equations and the low-Reynolds-number two-equation q - w turbulence model. The application to an annular cascade optimization demonstrated the validity, robustness and efficiency of this approach.

关 键 词：遗传算法 叶栅 叶轮机械 NAVIER-STOKES方程 逐次序列规划 响应面方法 SQP 粘性流动 数值模拟 

分 类 号：O35[理学—流体力学]