Data-Driven Identification of Lamb Wave Dynamics for Fatigue Monitoring of AL5754 Plates

This study investigates fatigue crack propagation in AL5754 aluminum plates using Lamb wave-based monitoring techniques. Multiple specimens were subjected to fatigue loading under varying force levels, R-ratios, and cycle counts, resulting in systematic crack growth. Lamb wave signals were acquired via surface-mounted piezoelectric disks and an arbitrary waveform generator at selected crack intervals. Signals were recorded using an oscilloscope and analyzed in both time and frequency domains via MATLAB and Python, revealing characteristic changes associated with damage progression. The tests were conducted using a custom experimental setup featuring a hydraulic actuator driven by pressure through an electronically controlled valve. The valve was modulated by a waveform generator to apply cyclic displacement at 5 Hz, while the resulting force was measured using a load cell connected to a National Instruments system interfaced with MATLAB. Beyond conventional signal analysis, curve fitting methods were employed to model the relationship between Lamb wave features and crack length. In addition, Sparse Identification of Nonlinear Dynamics (SINDy) was used as a data-driven approach to uncover the system’s underlying behavior. The results show that Lamb wave responses serve as sensitive indicators of crack growth, and that data-driven models provide strong predictive capabilities. This study supports the development of reliable in-situ structural health monitoring (SHM) methods for lightweight aluminum components and paves the way for future research involving machine learning frameworks such as Physics-Informed Neural Networks (PINNs).