流固耦合问题复杂，且非线性特征明显。求解该类问题时，解析法计算量小但要求很多简化，结果常无法代表真实的情况。数值法能给出量级上相当的估算，但其对计算机的要求较高且不同学者的研究结果常有所出入。解析法和数值法均属于正问题的范畴。本项目创新性地利用实测数据从反问题的角度开展研究。针对耦合问题中柔性结构非线性振动的动力特征，发展新的信号分析方法并将其与传统的时频分析方法进行对比。通过对实测数据的时频分析，研究结构振动频率和阻尼等随时间的变化情况。借助于理论分析和数值模拟等研究手段对现有的系统识别理论进行对比研究，探究适用于流固耦合系统的系统识别理论。结合分段线性化假定，对柔性结构流固耦合系统进行参数识别，探讨系统参数对振动特性的影响。基于时频分析和系统识别的结果，对现有的柔性结构有限元模型进行修正，使其更加合理。对流固耦合非线性振动问题进行时频分析、系统识别及模型修正具有重要学术意义和实用价值。

To study the dynamic mechanisms and characteristics of flexible structures during the fluid-structure interaction (FSI) is significant to the ocean engineering community. The problem of FSI is inherently complicated and of much nonlinearity. The researchers usually solve this problem analytically or numerically. However, it is crucial to take some assumptions to make the analytical solution available, which usually results in an analytical solution that contains some error. The numerical solution relies on the computational capability and different algorithms can gain different results. Either solving the FSI problem analytically or numerically belongs to the category of forward problems. For the first time, the applicants will study the FSI problem from the perspective of inverse problem. We will develop a new time-frequency analysis method, which decomposes signal in each window into complex exponentials based on state-space model. From the time-frequency analysis results, we can observe the varying frequencies. We will conduct a comparative study on some classical system identification methods and find which one/ones is/are suitable for FSI problems. As mentioned, the system of FSI is usually nonlinear. We will make a piece-wise linear assumption, and extract the system parameters from each segment. With this, we can find the connection between the factors of the structure or fluid and the system parameters. Because we can find the generalized mass, stiffness and damping matrices by system identification, hydrodynamic parameters, such as added mass, added stiffness and added damping, are available. Based on the study of time-frequency analysis and system identification, we will implement model updating on the finite element model by CMCM method. To conduct time-frequency analysis, system identification and model updating has great scientific and practical importance.