Bayesian continuous wavelet transform for time-varying damping identification of cables using full-field measurement

Cables serve as the primary load-bearing element in cable-stayed bridges, making their damping level critical for structural safety evaluation. Traditional operational modal analysis (OMA) faces challenges in damping identification due to result discreteness, and limited sensor deployment often leads to the loss of crucial modal information. This paper proposes a Bayesian continuous wavelet transform with Gabor wavelet (BCWT-G) method for time-varying damping identification using full-field measurement data. A computer vision technique combining the pyramid grafting network (PGNet) with neighboring frame pixel fitting (NFPF) is used to accurately capture full-field vibration data. The time-frequency domain properties of these data are then extracted and incorporated into a Bayesian probabilistic estimation framework for modal updating. The proposed method was validated through numerical simulations using a physics-based graphics model (PBGM), and actual cable testing under complex environments, demonstrating its effectiveness and robustness in identifying the time-varying dynamic characteristics of cables.