低雷诺数下翼型分离流动抽吸控制优化

Optimization of suction control on separation flow around an airfoil at low Reynolds number

为了获得不同目标下最优抽吸控制参数,开展了分离流动抽吸控制优化研究,基于RBF神经网络与遗传算法,发展了求解单目标和Pareto多目标问题的优化平台。针对NACA0012翼型表面分离流动,在其上表面设计了局部多孔分布式抽吸结构,将径向基函数(Radial Basis Function,RBF)神经网络作为CFD计算的代理模型,以减小计算量;采用遗传算法开展了单目标和Pareto多目标优化。优化结果表明:该优化设计平台具有良好的收敛性和准确度;以升阻比为单目标的优化使升阻比最大增加了2.4倍;Pareto多目标优化设计获得了分布均匀的、令人满意的Pareto解集,为设计者提供了一个可选的有效解数据库。在合理选择抽吸角度、抽吸孔径和孔间距的前提下,只需较小的抽吸系数,就可使升阻比获得明显增加,同时,还能保持较高的FOM值,使整个控制系统的能效比维持较高水平。

In order to obtain the optimal set of suction parameters for different objectives,an optimization method is developed to solve single-objective and multi-objective problems,by combining RBF neural network and genetic algorithm.Considering the separation flow over a NACA0012 airfoil surface,some local porous suction regions are mounted on the upper surface.RBF neural network is used as the surrogate model to substitute CFD computation,so as to reduce the amount of computation.The corresponding single-objective and Pareto multi-objective optimizations are performed using genetic algorithm.The optimization results show that the present optimization method has satisfactory convergence and accuracy.The maximum increase in liftto-drag ratios up to 2.4times is achieved after performing single-objective optimization of maximum lift-to-drag ratio.The uniform distributing and satisfied Pareto front is gained by Pareto multi-objective optimization,which provides a selective database with effective solutions.As long as the suction angle,hole diameter and hole spacing are reasonable,only smaller suction coefficient is required to get significant lift enhancement,and keep a high FOM value,so that the entire control system maintains a high level of energy efficiency ratio.