基于随机减量法的非平稳激励下模态参数识别

摘要

Aiming at the shortcoming of the existing modal parametric identification method with the assumption of stationary excitations,a modal parametric identification method under non-stationary excitation was presented here based on the radom decrement method.Firstly,the variance and the mean of the excitations were derived on the basis of the evolutionary spectral amplitude modulation theory for white noise amplitude modulation excitation,the related excitation with amplitude modulation and the non-stationary excitation with frequency and amplitude modulation.Then,the traditional random decrement method was expanded,and the modal parametric identification problem under non-stationary excitation was converted into a modal parametric identification problem based on free vibration response curves of structures.Finally,the modal parameters were identified using the eigen-system realization algorithms.The method was verified through a numerical model of a cantilevered beam and a two-story steel frame tests.The identification results showed that the proposed method can accurately identify natural frequencies and vibration modal shapes of the two structures under non-stationary excitation and the identification error for natural frequencies is less than 2%,it has a higher identification accuracy.%针对现有结构模态参数识别方法平稳激励假定的不足，提出了一种非平稳激励下的结构模态参数识别方法。以白噪声调幅激励、调幅相关激励和频率幅值调制共三类非平稳随机激励为对象，采用演化谱理论推导了随机激励分段叠加后的均值和方差表达式，然后在此基础上对传统的随机减量法进行拓展，将非平稳激励下模态参数识别问题转化为结构自由振动曲线的模态参数识别，最后利用特征实现算法识别结构的模态参数。通过一个悬臂梁数值模型和两层实验室钢框架进行验证，识别结果表明运用该算法可以准确识别出非平稳激励下系统的固有频率和振型，且固有频率的识别误差在2％以内，具有较高的识别精度。