中文摘要：

POD？Galerkin 方法是构建非定常流动低阶模型的有效方法。然而，研究表明基于周期振荡样本及其时均解构建的低阶模型只能反演流场振荡饱和后的周期性运动，不能复现流动从不稳定定常解开始振荡的发散过程，不便于流动的稳定性分析和控制研究。通过尝试选择流场进入周期性运动之前的样本来构建低阶的流体动力学模型，并用于反演流动失稳初期微幅振荡的流场。以低雷诺数（Re＝100）下圆柱绕流为例，构建了卡门涡街失稳初期的流动降阶模型。通过与 CFD 数值结果对比，表明选择合适的样本数据，构建的低阶模型可以复现流场发散初期的频率和阻尼等特性。

英文摘要：

PODGalerkin method is an effective way to build lowdimensional models for unsteady flows. However, studies have shown that lowdimensional models, based on snapshots of periodic oscillations and the time averaged solutions, can only rebuild the flow field of saturated periodic oscillating state, but iftcannot rebuild the divergence process of the flow from unstable steady solution to final unsteady saturated oscillation state. So it is inconvenient for stability analysis of the flow and for the controller design in future research. In this paper, we attempt to found lowdimensional dynamics modelsby using the snapshots of the fluid before entering the cyclical movement. And then we use this model to rebuild the flow field of the initial development stage with small oscillating amplitude.We take the flow past a circular cylinder at a low Reynolds number of 100 as an example to build a lowdimensional model for the initial instable stage of Karman vortex street. It is shown that, by choosing appropriate snapshots, the reducedorder model can rebuild some characteristics of the initial divergence regime of the flow, such as frequency and damping characteristics. This study, we believe, establishes a good foundation for the fluidstructure coupling analysis and flow instability control in future research.