Stiffness state evaluation of high-speed railway bridges based on spectral features of PCA-fused vehicle responses

Indirect bridge assessment based on vehicle responses provides a promising approach for rapid evaluation of high-speed railway bridges. However, under high-speed conditions, the short bridge-crossing duration limits the effective frequency resolution of vehicle response spectra, and vehicle responses are also influenced by vehicle natural frequencies, vehicle passing frequencies, track irregularity, and noise. The combined effects of limited frequency resolution and spectral interference make it difficult to stably extract bridge-frequency-related spectral features from a single response channel. To address this issue, considering the non-negligible vehicle and wheelset masses in railway vehicle-bridge interaction, a two-degree-of-freedom railway vehicle-bridge interaction model incorporating the vehicle body, bogie, wheelset mass, primary suspension, and secondary suspension is established. Subsequently, based on the bridge-frequency-related components contained in both the vehicle body and bogie responses during bridge crossing, this study proposes a bridge stiffness state evaluation method using the local peak-frequency shift and Pearson spectral correlation coefficient (PSCC) of principal component analysis (PCA)-fused vehicle responses to assess different bridge stiffness states under high-speed conditions. Numerical results for a representative 32 m simply supported bridge demonstrate that PCA-fused responses provide more stable bridge-frequency-related spectral features under high-speed conditions. PSCC reflects slight bridge stiffness degradation before noticeable local peak-frequency shifts occur, whereas local peak-frequency shifts become evident under larger stiffness degradation, enabling bridge stiffness state evaluation without requiring direct identification of bridge natural frequencies. The proposed method remains robust under variations in vehicle mass, suspension stiffness, vehicle speed, track irregularity, and noise, demonstrating its feasibility for high-speed railway bridge stiffness state evaluation.