Abstract:Measured output signals are inevitably contaminated with noise. A low rank approximation of a Hankel matrix method was proposed to remove noise from the measured data, and then modal parameters were estimated from the filtered data. In the method, a Hankel matrix was constructed based on the impulse response function and then the singular value decomposition (SVD) was used to obtain the relative change ratio of the decreasing singular values. Theoretically the maximum change ratio corresponds to the model order. An iterative low rank approximation method was applied to get the filtered signals. The effects of chosen size of a Hankel matrix on the performance of the noise reduction and computational time were investigated. The effectiveness of the proposed method was verified by using measured signals from model test of a cantilever beam.