Damage identification methodology for composites via adaptive mode selection and unbiased weighted curvature

In conventional vibration-based damage identification for composites, the subtle local features caused by damage are easily obscured by dominant global structural dynamics and measurement noise. To address these practical challenges, this paper proposes a robust baseline-dependent damage identification methodology built upon two integrated innovations: (1) a Feature Prominence Index (FPI) that quantitatively identifies the most damage-sensitive vibration modes, eliminating subjective mode selection; and (2) an unbiased Mode Shape Weighted Square of Curvature Difference (MSW-SCD) index that amplifies localized damage signatures while suppressing noise in low-confidence regions. The MSW-SCD achieves this by decoupling the damage-induced curvature perturbation from the healthy curvature field and employing modal amplitude as a spatial confidence weighting factor. In addition to these core contributions, an image-based preprocessing strategy is employed to ensure stable modal curvature computation, so that reliable inputs are obtained for the FPI and the MSW-SCD index. The methodology's effectiveness is rigorously validated through numerical simulations and experiments. Under simulated high-noise conditions, it demonstrates precise identification of both single and multiple damage. Experimental validation confirms the methodology's practical effectiveness, with its damage identification results showing strong agreement with ultrasonic C-scan references. By systematically overcoming the key practical barriers of measurement noise and damage feature ambiguity, this work advances modal curvature theory from a theoretically sensitive concept into a reliable, objective tool for real-world structural health monitoring of composite structures.