• Earthquakes and Structures
• /
• v.27 no.5
• /
• pp.419-429
• /
• 2024

The safety and stability of bridges are critical to traffic safety. However, post-earthquake ground motion records often contain noise, which undermines the accuracy of seismic response analysis in bridge structures. To tackle this issue, we introduce a method that optimizes Variational Mode Decomposition (VMD) parameters using the Sparrow Search Algorithm (SSA) and combines it with Wavelet Thresholding (WT) to eliminate noise from strong motion signals. SSA is employed to identify the optimal VMD parameters [K, α], followed by the selection of effective modes based on the Variance Contribution Rate (VCR). These modes are then subjected to WT noise reduction, resulting in a high-quality reconstructed strong motion record. The method was validated using both simulated signals and ground motion records. In simulations, it demonstrated a 31.35% reduction in Root Mean Square Error (RMSE), a 31.6% decrease in the Smoothness Indicator (R), and a 1.17% improvement in the Correlation Coefficient (CC), compared to other methods. For ground motion records, it more accurately preserved seismic features than traditional wavelet denoising. When applied to the seismic response analysis of the Dahejia Bridge during the Jishishan earthquake, the denoised ground motion records obtained by this method produced force predictions on pier bearings that closely matched the field-observed damage, outperforming predictions based on traditional wavelet denoising. These findings confirm the accuracy and practicality of the proposed method.