ABSTRACT
This paper presented a study of operational modal analysis and cable tension estimation using ambient vibration in long-term bridge SHM. Bayesian operational modal analysis under ambient vibration was conducted in a cable-stayed bridge by the means of a Bayesian FFT and the coupling effect of cable-deck vibration is discussed. A Bayesian cable tension estimation method is proposed with the identification uncertainty in the estimates of cable tension. Further, the effect of EOVs in long-term SHM of cable tension was investigated with an integration of identification uncertainty using a probabilistic predictive model. The result indicated the effect of EOVs in the longest cable is much more dominant compared to the shortest cable at bridge deck end. Finally, a Gaussian mixture model (GMM) based anomaly detection method in the long-term SHM of stay cable was proposed and investigated with the likelihood as a damage sensitive feature (DSF) and the damage effect according to a simulated analysis. The result showed that this method can provide a meaningful indication of damage under a certain premise.
