Frequency identification of stay cables using the Bayesian spectral density approach with uncertainty quantification and temperature effects

Abstract To mitigate the effects of environmental disturbances and measurement noise affecting the frequency identification of stay cables in cable-stayed bridges, this study enhances a Bayesian spectral density approach to jointly identify modal frequencies and their probabilistic uncertainties from single-sensor measurements, and validates the method using two years of monitoring data from the Qingzhou Channel Bridge. The first ten modal frequencies of the L cable (longest stay cable) and the S cable (shortest stay cable) in January and July were identified, and their uncertainties evaluated. Additionally, the correlation between cable frequencies and temperature variations was analyzed. Results indicate that the frequency estimates exhibited coefficients of variation within 0.30% for the L cable and within 1.30% for the S cable. Pronounced seasonal variations were observed, with the L-cable frequencies averaging 0.32% higher in January than in July, while the S-cable frequencies were 0.72% lower. Temperature increases also magnified uncertainty for the S cable, raising its July standard deviation and coefficient of variation by 27% and 56%, respectively. Temperature-frequency correlation analyses revealed a persistent negative correlation for the L cable—although the sensitivity of its higher-order modes diminishes in July—whereas the S cable changes from a negative correlation in January to a positive correlation (up to 0.76) in July. These results provide valuable insights into the seasonal and thermal sensitivity of stay cable dynamics and demonstrate the applicability of the proposed method to long-term field monitoring.