Vibration monitoring method for long-span bridges based on distributed optical fiber acoustic sensing

The distributed optical fiber acoustic sensing (DAS) technique has been regarded as an effective approach to monitor the dynamical response of long-span bridges. However, most signal processing methods for DAS were designed for perturbation detection. Inadequate consideration of fading noise and error propagation led to compromised frequency estimates and distorted mode shapes. In this study, a vibration monitoring method using a DAS system is proposed for long-span bridges. This method combines optical signal demodulation and dynamics principles to improve accuracy and reliability. We develop a novel untwisting algorithm to simultaneously suppress polarization and interference fading noise, as well as disturbance-induced waveform distortions. Considering the characteristics of optical phase time histories, phase modal parameters calculated through the eigen-system realization algorithm are optimized iteratively to identify real modes. A numerical simulation of bridge monitoring validated the feasibility of the proposed method under full-scale conditions. A vibration experiment was conducted on a 6.5 m model of a cable-stayed bridge. Phase time histories of 25 sensing points could be retrieved with clear waveforms, the average signal-to-noise ratio was improved to 14.52 dB. The first 7 natural frequencies were identified, which agreed well with the results of fiber Bragg gratings. Four phase mode shapes were extracted with an average modal assurance criterion of 0.9406. We achieve distributed frequency and mode shape identification with 0.25 m measuring resolution, which is crucial for the condition assessment of long-span bridges. The present study can serve as the basis for further application of DAS in bridge health monitoring.