Intelligent Model for Flaws and Defects Classification and Identification in Adhesively Bonded FRP Joint

Adhesively bonded fiber-reinforced plastic (FRP) joints are widely used in engineering applications due to their lightweight, high strength, and corrosion resistance. However, flaws and defects such as voids, delamination, and micro-cracks compromise their structural reliability, requiring effective detection and classification methods. This study presents an intelligent, data-driven approach for flaw identification in adhesively bonded FRP joints using the K-means clustering algorithm. Acoustic Emission (AE) signals were employed as input features, enabling unsupervised classification of defects without prior labeling. The model performance was validated using multiple metrics, including the Silhouette Score (0.46) and the Calinski-Harabasz Index (90.72), which confirmed good cluster compactness and separation. Results show that the method successfully categorized defects into three severity levels, facilitating accurate identification of critical fault zones. The proposed approach offers a reliable, non-destructive tool for structural health monitoring, with potential applications in preventive maintenance, safety assurance, and performance optimization of FRP-based structures.