从响应信号辨识斜拉桥模型的模态参数

针对斜拉桥模型在平稳随机激励下的模态参数辨识问题，研究了提高子空间方法参数辨识精度和可信度的有效途径。在矩阵奇异值分解的基础上，文中使用信号特征分量的能量指标来估计模型阶次，同时给出一种特征频率随拟合数据变化的稳定图。在虚假特征显示能力以及降低数值运算量等方面，这种形式的稳定图与能量指标的结合具有正更大的优越性。在斜拉桥模型的模态参数辨识中，物理模态和寄生模态得到了很好的分离，而且斜拉桥模型在分析频带内的物理模态被全部识别出来。辨识结果的比较说明了给出的辨识方法是有效的。

Modal Rarameter Identification of a Cable-stayed Bridge Model by Merely Using Response Data

Focusing on the parameter identification of a cable-stayed bridge model under stationary stochastic excitation,an improved subspace identification method is presented.On the basis of singular value decomposition,an index referred to as CEI is introduced to estimate model order and an alternative stabilization diagram that reflects the variation of estimated frequencies with data increment is also given to reliably separate spurious estimates and to reduce computation complexity.In the identification of bridge modal parameters,spurious and physical estimates are clearly indicated and all modes within the analysis frequency band are identified,which demonstrates that the alternative stabilization diagram is superior to the conventional one in terms of the accuracy and completeness of identification.