Executive Summary : | Damage detection is a crucial aspect of structural health monitoring, particularly in vibration-based systems like bridges. Vibration signals from sensors are used in system identification strategies to identify properties like natural frequency, damping, and stiffness. However, these methods require validated analytical or finite element models, which can be challenging for large-scale infrastructure. To overcome this, vibration signal processing methods are proposed, which include time-, frequency-, and time-frequency domain methods. Time domain methods involve time series models, while frequency domain methods based on the Fourier transform involve bulk processing, leading to averaging of spectral content. Time-frequency methods based on wavelets and short-time Fourier transform have been proposed, but they can alter signal features and averaging over a shorter duration. This work proposes an instantaneous spectral analysis of vibration signals using the single frequency filtering (sFF) method. This allows for visualization of instantaneous variations in spectral content with better resolution of higher modes, which may indicate early damage. The proposed work will be carried out in three phases: tests on beam specimens to acquire vibration signals, sFF analysis to relate signal features and system properties, studying the influence of signal features on changes in system properties by varying mass and/or stiffness of specimens, and finally, presenting a framework for damage detection using instantaneous spectral analysis for scaling to field studies. This approach could contribute to preemptive solutions in structural health monitoring, addressing the challenges faced by old and new infrastructure. |