APPLICATION OF SECOND-ORDER STATISTICS BLIND IDENTIFICATION ON IDENTIFYING MODAL PARAMETERS

It is important to identify structural modal parameters in time and accurately for accurate damage identification and health monitoring of structures. A new method of identifying the structural modal parameters is developed in this paper. This method is based on the AMUSE algorithm in blind source separation theory. It considers the free responses or pulse responses of an oscillatory system as the analysis object and increased virtual measured points by using Hilbert transformation to the data. This method uses a variance covariance matrix with different time lags to build joint matrices. It realizes identifying the structural modal parameters by solving the eigen value of joint matrices with different time lags. Introducing the joint matrices conquers the instability coming from the AMUSE algorithm, which only uses two covariance matrices. A set of applications have been performed. The numerical results show that the method developed in this paper has the following advantages: the calculation is simple; the identification has high accuracy, not limited by the choice of time lags, not sensitive to measure white noise and having good robustness.