Identification of bridge modal parameters from GNSS data by integrating IEWT and robust ICA algorithm

Empirical wavelet transform is often used to process global navigation satellite system (GNSS) bridge deformation monitoring data, but it leads to inaccurate band division and too many spurious modes. To address these problems, this study proposes an improved algorithm to identify bridge modal parameters, which combines improved empirical wavelet transform and robust independent component analysis (ICA). The proposed method adopts the autoregressive power spectrum of an improved covariance algorithm, instead of the Fourier spectrum, for band division. Additionally, it performs noise reduction and reconstruction of multi-channel GNSS monitoring data. The reconstructed signal is inputted as multi-channel observation signal into robust ICA to extract features of the source signal. Finally, the natural excitation technique and Hilbert transform are used to solve the self-oscillation frequency and damping ratio of the structure. The proposed method is validated using both simulation data and the GNSS monitoring data of the Wilford suspension bridge. The results show that the proposed method can effectively reduce the measurement noise and successfully identify the first-order vibration frequencies and damping ratios of bridge. This algorithm can also be applied in the parameter identification of other engineering structures from GNSS data.