Vibration-Based Numerical and Experimental Analysis on Delamination Damage Identification for Composite Laminates

Composite laminates possess excellent specific mechanical performance and are widely adopted in aerospace, rail transit and national defense industries, whereas interlayer delamination, as a typical concealed damage, seriously threatens structural safety. Most existing vibration-based delamination detection algorithms rely on intact structure baseline data, which restricts their practical engineering application. Aiming at this limitation, this paper investigates two baseline-free delamination identification approaches combining modal shape curvature and uniform load surface (ULS) curvature for eight-layer symmetric cross-ply composite laminates. Finite-element models containing boundary-layer and middle-layer delamination are established using the node-merging method. Modal shape curvature and ULS curvature are computed via the central difference method, and two baseline-free schemes, namely the surface smoothing algorithm and curvature mode change rate method, are implemented for damage feature extraction. Three quantitative evaluation indices, including Peak to Background Ratio (PBR), localization error and Signal to Noise Ratio (SNR), are introduced to objectively compare the identification performance of different curvature indicators and algorithms. Numerical results demonstrate that both methods can effectively locate delamination under 3% random measurement noise. Boundary-layer delamination yields more prominent damage features than middle-layer delamination; modal shape curvature outperforms ULS curvature with 1.29–1.53 times higher PBR and 1.23–1.52 times higher SNR, alongside reduced localization error. Under identical conditions, the curvature mode change rate method achieves 1.51–1.99 times higher PBR, 1.22–1.36 times higher SNR and localization error reduced to 0.50–0.79 of that obtained by the surface smoothing algorithm. Scanning laser Doppler vibration tests on carbon-fiber specimens with preset delamination defects further validate the proposed methods. Experimental results reveal that under actual measurement noise, the curvature mode change rate method attains maximum PBR of 12.56, localization error down to 6.46 mm and SNR up to 14.13 dB, exhibiting better comprehensive performance compared with the surface smoothing algorithm. This work provides quantitative reference for baseline-free delamination detection of composite laminates and offers a feasible technical route for practical non-destructive inspection.