基于盲源分离的结构模态参数识别

Method of modal parameters identification based on blind sources separation

提出一种结构模态参数识别的新方法。该方法以盲源分离理论中基于二阶统计量的AMUSE算法为基础,通过对测试数据Hilbert变换构建分析数据矩阵,通过求解不同时滞下数据协方差矩阵的广义特征值问题实现对结构模态参数的识别。数值算例结果表明,该方法不仅适用于实模态情况,同时适用于复模态情况,且计算简单,识别精度高,对测量白噪声有很好的鲁棒性。

A new method for structural modal parameters identification was developed.Based on the AMUSE algorithm of second order statistic in the blind source separation.Hilbert transformation of the measured data was used to build analysis data matrices and the structural modal parameters identification was realized by solving the generalized eigenvalue problem of the variance and covariance matrices under different time lags.The results of numerical applications show that the method remains valid not only for the real modes situation but also for the complex modes situation.Further more,the method is of the advantages of calculation simplicity,high identification accuracy and excellent robustness for measuring white noises.