Abstract: Based on wavelet multi-resolution analytical theory, differential equation of linear structural dynamic system was decomposed and dynamic time series were described in different scales. The loads and structural responses were transformed by orthogonal wavelet and the parameter identification equation was gained by wavelet coefficients on different frequency bands. The physical parameters were obtained by weighting least-square method. The weighing values were derived by the ratio of noise to signal on different frequency band. The numerical simulations show that the proposed method can improve the identification accuracy greatly and the accuracy is raised by the level of wavelet decomposition.