中文摘要：

对流体使用降阶模型是流固耦合计算中为提高计算效率常用的手段之一,但采用降阶模型对模型维数的降低,或全阶模型数据与降阶模型方程的不一致性会影响降阶模型的准确性.本文引入人工涡常量和遗传算法对流固耦合问题中降阶模型进行校正研究,首先基于伽辽金投影法和本征正交分解法投射到最主要本征模态空间上,得到流体的降阶模型;然后引入人工“涡”对降阶模型的系数进行校正,最后采用遗传算法对校正模型中的系数进行估计.将本文提出的校正降阶模型应用于典型流固耦合问题分析中,对比了校正前后位移和力的变化,以及校正前后降阶模型的误差变化,结果表明校正后的降阶模型的计算准确性和效率大大提高,证实了降阶模型校正的必要性和本文降阶模型校正方法的有效性.

英文摘要：

Fluid reduced order models （ROM） are commonly used in fluid-structure interaction. However, the accuracy of reduced order models would be affected due to dimension reduction of reduced order models or inconsistency of full-order model and reduced order model. Study is performed on calibration of reduced order models combining artificial vortex with genetic algorithm. Fluid equations are projected onto major engin-modal space using Galerkin projection method and proper orthogonal decomposition to obtain fluid ROM. Calibration is performed by introducing artificial eddy to calibrate the coefficients of the ROM. Genetic algorithms are employed to estimate the coefficients. The proposed calibrated ROM is applied to the typical fluid-structure interaction computation. Displacement, force and errors before and after calibration are compared respectively. The results show that accuracy and efficiency are greatly improved with calibrated ROM, which validates the calibrated ROM.