Multiscale Time Series Decomposition for Structural Dynamic Properties: Long-Term Trend and Ambient Interference

In structural health monitoring (SHM), variations of dynamic properties are paramount to indicate the status structures. Structural time variant due various long-term effects (e.g., deterioration) and periodic temperature, humidity, traffic). Sometimes will interfere with quantification effects. Though important, there still limited research studies aiming distinguish these two Given amount SHM data, it is possible solve issue from a data-driven perspective. This article proposes series decomposition methodology divide into parts, multiscale holiday error parts. We extract 10-year long-span bridge using fast Bayesian FFT identification algorithm choose modes that as examples explain how our proposed works. use parts rules deterioration. The utilized find relationships periodically varying ambient conditions (i.e., temperature humidity) in different scales yearly, weekly, daily). Then, can be distinguished. For effects, modal frequencies tend decrease but damping ratios seem increase. we increment lead both ratios. induced by deterioration same amplitudes. not significantly related humidity. could provide references for damage detection safety assessment similar bridges.