Simulation and experiment of a blind subspace identification method

A covariance-driven subspace identification method is presented to identify weakly excited modes. In this method, the traditional block Hankel matrix is reshaped to enhance identifiability of weak characteristics. The robustness of eigenparameter estimation to noise contamination is reinforced by the reshaped block Hankel matrix. An alternative stabilization diagram in combination with component energy index (CEI), which indicates vibration intensity of signal components, is adopted to separate spurious and physical modes. Simulation of a vibration system of multiple-degree-of-freedom and experiment of a metallic frame structure subject to wind excitation are presented to demonstrate efficacy of the proposed blind method. The performance of this blind method is assessed in terms of its capability in extracting weak modes as well as the accuracy of estimated parameters. The results have shown that the proposed blind method gives a better estimation of weak modes from response signals of small SNR and gives a reliable separation of spurious and physical estimates.