Operational Modal Analysis Method for Long-Span Suspension Bridges Based on Data Fusion

Subjected to the influence of dense and false modes, it is difficult to continuously track the accurate structural modal parameters for long-span suspension bridges. To improve the robustness of the structural Operational Modal Analysis (OMA), a method combining the Multi-rate Kalman Filter (MKF) and the Covariance-driven Stochastic Subspace Identification (SSI-COY) is proposed from the perspective of optimizing the analyzed object. Firstly, based on the double integral relationship between the girder's acceleration and displacement responses, a state space equation is constructed to characterize the vibration of the main girder, where the fusion weight ratio between the two data types is determined based on the energy of the first peak in the self-power spectral density. In this way, the displacement response is recursively reconstructed by the MKF data fusion. Then the SSI method is used to identify the measured modal parameters from the numerous vibration time histories during different periods. The identification results are screened by the simulated vibration modes and the modal assurance criterion (MAC). Taking the 1-hour measured responses of a 660m-long suspension bridge within two weeks as the research object, these corresponding 168 sets of SSI identification results from the original acceleration signals, the original displacement signals, and the MKF reconstruction responses are statistically compared, indicating that the data fusion method can significantly improve the OMA accuracy and robustness for the long-span suspension bridge.