Role of Ultrasonic Guided Wave-Based SHM in Detecting Low-Velocity Impact Damage in CFRP Laminates

Ultrasonic Guided Waves (UGWs) are widely used in Structural Health Monitoring (SHM) systems for their ability to inspect large areas with minimal sensor intrusiveness. However, extracting reliable damage information from UGW signals remains challenging, especially in complex composite structures. This study presents a data-driven methodology for classifying low-velocity impact damage in Carbon Fiber Reinforced Polymer (CFRP) panels using UGWs and unsupervised clustering. Ten CFRP plates, each equipped with six piezoelectric transducers, were subjected to controlled 15 J impacts at their geometric centres. For each panel, UGWs were acquired in both pristine and damaged conditions through a sequential pitch-catch configuration. A set of 32 scalar features was extracted from each signal, and all possible pairwise and triplet combinations of features were explored to evaluate their effectiveness in damage classification. Unsupervised clustering algorithms were applied to assess the separability of healthy and damaged states, with performance evaluated via silhouette score, purity, and cluster balance. The results highlighted the ability of specific feature combinations to consistently differentiate impact damage without prior labelling, offering a lightweight and interpretable approach for structural diagnostics. This work demonstrates the viability of combining UGWs with unsupervised learning for effective and scalable SHM of composite structures.