一种代理遗传算法及其在气动优化设计中的应用  

An Effective Surrogate-Assisted Genetic Algorithm for Airfoil Aerodynamic Optimization Design

作　　者：苏伟[1] 高正红[1] 夏露[1] 

机构地区：[1]西北工业大学航空学院,陕西西安710072

出　　处：《西北工业大学学报》2008年第3期303-307,共5页Journal of Northwestern Polytechnical University

基　　金：国家自然科学基金(10502043)资助

摘　　要：遗传算法具有良好的鲁棒性和全局优化等优点,但是需要进行大量目标特性的计算,因此计算量很大。针对这一问题,利用代理模型计算量小的优点,将代理模型引入到遗传算法中,建立了一种高效的代理遗传算法。在该算法中,以遗传算法为整体框架,在优化搜索中部分使用代理模型进行目标特性分析,大大减少了计算量。为防止代理模型不精确带来的影响,在优化过程中通过引入EI方法,较好地解决了算法中校正个体的选择问题。为了验证方法的有效性,使用该算法进行了翼型的气动外形优化设计,升阻比提高了40%。与基本遗传算法相比,该算法的优化结果与之相当,但计算时间减少了约75%。结果表明该算法对遗传算法的改进是有效的,适合进行气动外形优化设计。We now present SAGA （Surrogate-Assisted Genetic Algorithm）, an effective algorithm for airfoil aerodynamic optimization design. In the full paper, we explain SAGA and its effectiveness in airfoil aerodynamic design. In this abstract, we just add some pertinent remarks to the three topics of explanation. The first topic is. Kriging model used in SAGA. This model is taken from Ref 2 by J. Sack et al. The second topic is. SAGA. In the second topic, most of the individuals are evaluated by the timesaving surrogate model. Also in the second topic, we take the EI（Expected improvement） method in Ref 4 by R. J. Donald et al to select the calibration individuals effectively. The third topic is. airfoil aerodynamic optimization design. In the third topic, we take RAE2822 airfoil and optimize it with SAGA and with SGA（Simple Genetic Algorithm） respectively; the results, given in Table 1 in the full paper, show preliminarily that both SAGA and SGA can increase through optimization the lift-drag ratio of RAE 2822 airfoil by about 40%. Also in the third topic, we point out that the optimization time required by SAGA is only about 25% of that of SGA.

关 键 词：代理模型 遗传算法 代理遗传算法 EI方法 气动优化设计 

分 类 号：V260[航空宇航科学与技术—航空宇航制造工程]