基于模拟退火算法与响应面模型的三维气动优化设计方法

Three-dimensional aerodynamic optimization design method using simulated annealing and response surface model

提出了一套适用于高耗时三维气动设计问题的优化设计体系。其主要思想是采用改进拉丁超立方体试验设计选取样本点,应用自行开发的三维粘性流场求解程序进行流场计算建立数据库,采用二次响应面方法建立近似模型,再应用高效模拟退火算法进行全局寻优。以NASA rotor67为对象,在详细进行流场计算基础上采用所提出的优化体系对其进行了三维积叠优化设计。在对流量、效率加以严格约束的条件下,总压比可提高1.8%,对流场结构进行了分析。优化结果表明本优化方法省时,适于三维气动设计的特点。

This paper presents an optimization design system suitable for computationally expensive three-dimensional aerodynamic design problems. The improved Latin hypercube design of experiment was used to select sample points which were then analyzed by a three-dimensional Navier-Stokes solver developed by authors to construct database. The approximation model was established by the quadratic response surface method. Then the model was optimized by using effective simulated annealing algorithm to obtain the optimum design. To demonstrate its performance, the optimization redesign was performed on NASA rotor 67 transonic fan rotor blades with three-dimensional leading-edge stacking line. As a result of the optimization, the pressure ratio was successfully increased by 1.8 %, as well as to limit little mass flow and adiabatic change. Then a detailed study of flow structure was presented and the results have proven that this design optimization method can help turbomachinery designers to design higher-performance machines with shorter design cycle and lower design costs.