New Spectro-Temporal Fusion Framework Enhances Structural Health Monitoring Through Vibration Analysis

A new framework known as Hybrid Spectro-Temporal Fusion has been created to enhance structural health monitoring by examining vibration responses from engineering systems. This method combines arrival-time interval descriptors with spectral characteristics to effectively capture both detailed and broad vibration dynamics. Experiments conducted with data from an LDS V406 electrodynamic shaker demonstrate that these spectro-temporal representations surpass traditional input methods. The study reveals that a temporal resolution (Δτ) of 0.008 benefits conventional machine learning models, whereas a finer resolution of 0.008 maximizes the capabilities of deep learning architectures. In addition to classification accuracy, the framework features an extensive stability analysis based on condensed indices, and introduces the Spectro-Temporal Alignment framework as part of this innovative approach. Structural health monitoring is vital for ensuring safety through vibration analysis.