Gaussian Process Regression Prediction Model for Vortex-Induced Vibration of Suspension Bridges Driven by Real Bridge Monitoring Data

Vortex-induced vibration (VIV) is a great threat to the safety of vehicles traveling on large-span bridges, and has a certain impact on the structural safety and durability of bridges. It is very important to predict and warn of VIV events in advance. In this paper, the continuous monitoring data of Xihoumen Bridge over the last 4 years were used to identify and intercept 67 VIV events, including their formation, stable oscillation, and subsequent decay phases, as recorded in acceleration segments. Samples with clear labels were obtained using sliding windows, which were complemented by other VIV characteristic data to construct a sample database related to historical VIV events. A VIV prediction model based on the VIV samples was established and trained using Gaussian process regression (GPR). The model realized the dynamic prediction and perception of VIV events, and the validity of the model was verified using two VIV events of real bridges with different development levels.