Bridge cables vibration frequency identification and fatigue life prediction based on SHM data with machine learning

Cables are an essential component of bridge structures, and their safety is directly related to the service life of the bridge. In particular, the change in cable force is a crucial indicator for assessing the safety of the cable. In this paper, the acceleration signal of the cable structure is obtained using a structural health monitoring (SHM) system, and the signal is transformed into the time-domain frequency using the Fourier transform method (FFT). Then, the frequency-cable force equation is established to quickly and accurately obtain the change in cable force in the time domain. Finally, the statistics of the cable force range are obtained using the rainflow counting method. This method enables the rapid and accurate assessment of the fatigue life of the cable structure when physical parameters of the cable are available. In particular, a probability prediction model is constructed based on the Bayesian emulator, and the influence of three covariance functions (Squared Exponential (SE), Matern-3/2 (MA-3/2), and Matern-5/2 (MA-5/2)) on the prediction performance is discussed, which is verified using the SHM data. The results indicate that for acceleration signal data with significant environmental noise, the Bayesian model based on MA-3/2 covariance performs the best in terms of prediction. However, when the acceleration signal data is relatively smooth, the prediction performance of the Bayesian model based on the three covariance functions is the same. At the same time, by combining on-site monitoring data, it can be inferred that the fatigue life of the bridge stay cables is greater than the service life, meeting the safety requirements for bridge operation and maintenance.