一种数据驱动的翼型流动转捩预测方法

研究翼型绕流的转捩预测方法，对于翼型流动细节的精确模拟和气动力的准确计算以及精细化设计均具有十分重要的意义.采用动模态分解（dynamic mode decomposition，DMD）代替线性稳定性理论（linear stability theory，LST）与eN方法结合，不需要求解稳定性方程，成为一种数据驱动的翼型边界层转捩预测新方法，称为DMD/eN方法.在原有方法的基础上，改进了DMD网格线生成方法和扰动放大N因子的积分策略，并将RANS求解器与改进的DMD/eN方法进行耦合，实现了翼型定常绕流转捩预测自动化.采用该方法对LSC72613跨声速自然层流翼型以及NLF0416低速自然层流翼型在不同攻角下的绕流进行转捩预测，转捩点计算结果均与实验值和LST/eN方法吻合良好.该方法计算得到的N值增长曲线与LST/eN方法的包络线也较为吻合，进一步验证了积分策略的正确性.改进的DMD/eN方法可作为自然层流翼型设计的新的有力工具. 

A Data-Driven Transition Prediction Method for Flows over Airfoils

Transition prediction method is crucial for simulation of flow over airfoils, since it can improve the accuracy of capturing of flow phenomena and predicting the aerodynamic forces, as well as the refined design. It is a novel data-driven transition prediction method of airfoil boundary layer by use of dynamic mode decomposition(DMD) instead of linear stabi-lity theory(LST) combining with eN method, referred to as DMD/eN method. Based on this method, this paper improved the DMD grid line generation method and N factor integration strategy. The improved DMD/eN method was coupled into the RANS solver, which achieved the automation of transiton prediction for steady flows over airfoils. Transition prediction of flows around LSC72613 airfoil and NLF0416 airfoil at various angles of attack were carried out. The predicted transition locations were in reasonably good agreement with the experimental data and the results of LST/eN method. By comparison, the calculated growth curve of N factor also accords with the envelop of LST/eN method well, which further demonstrates the feasibility of the developed integration strategy. This will be a new powerful tool for design of NLF airfoils.