Signal denoising for timber bridge structural health monitoring using optimized variational mode decomposition

Operational and measurement noise presents a major challenge to the accuracy and reliability of structural health monitoring (SHM) systems, particularly in timber bridge applications. This study investigates the effectiveness of Variational Mode Decomposition (VMD), an advanced signal processing technique, in enhancing the quality of response measurements from a laboratory-scale pedestrian timber bridge. Acceleration and strain signals were collected under both intact and damaged conditions to perform a detailed signal analysis. VMD was applied to decompose the signals into narrowband intrinsic mode functions (IMFs), enabling the isolation of structural responses from noise. The method yielded high cross-correlation values (above 0.998) between original and reconstructed signals, confirming that critical features were preserved. Furthermore, energy retention analysis across IMFs revealed distinct patterns reflective of structural condition, with meaningful content concentrated in the lower-order modes and noise primarily captured by the final components. These findings confirm the potential of VMD as a robust preprocessing tool for noise reduction within SHM frameworks, supporting improved interpretation of structural responses in timber bridge structures.