Enhanced quantitative damage detection for composite laminates based on compressed features of Lamb waves

To achieve a comprehensive and high-fidelity evaluation encompassing the initiation, localization, morphology, and geometric dimension of damage in composite laminates, an advanced quantitative damage detection framework is proposed by seamlessly integrating a multi-domain feature compression model with an enhanced spatial imaging algorithm. A task-specific channel attention network, optimized via a squeeze-and-excitation mechanism, is developed to compress multi-domain Lamb-wave features into a compact, damage-sensitive representation. On this basis, an Enhanced Damage Index (EDI) with inherently robust immunity to environmental perturbations is established to facilitate holistic structural health monitoring. Utilizing the path-specific EDI values derived from a distributed sensor array, an EDI spatial probability fusion imaging algorithm is executed to precisely isolate the most probable center of the defect zone along with its associated wave-scattering trajectories. To suppress artifacts arising from spurious edge reflections, an adaptive density-based spatial clustering algorithm is introduced to effectively filter true boundary-scattering coordinates from statistical outliers. The macroscopic topology of the internal degradation is subsequently reconstructed by shape-matching the convex hull of the isolated scattering cluster using dynamic time warping against preset geometric templates. Ultimately, the definitive damage center and quantitative boundary scale are calibrated via the maximum inscribed regular shape tightly bounded by the convex envelope. Extensive validation conducted on both numerical finite-element models and physical experimental platforms yields remarkable agreement between the predicted profiles and actual damage states. The results robustly confirm the applicability, exceptional localization precision, and quantitative reliability of the proposed framework for multi-parameter defect characterization in anisotropic composite structures.