Laser-Ultrasonic Thickness Gauging of Non-Ideal Thin Plates Using a Multi-Feature ZGV Calibration Framework
Yuebin He, Maodan Yuan, Xuanrong JiLaser-ultrasonic zero-group-velocity (ZGV) Lamb-wave thickness gauging is examined for uniform, curved, rough-surface, and variable-thickness 7050 aluminium plates. Thickness is inferred from the first-order symmetric (S1) ZGV resonance using a material-specific frequency–thickness constant; a separate velocity measurement is unnecessary for each estimate after calibration, but dependence on elastic properties remains. Resonance frequencies are extracted using a Hann-windowed, zero-padded fast Fourier transform (FFT). Spectral entropy H and instantaneous-frequency standard deviation σf are investigated as inputs to empirical within-group corrections in separately labelled specimen groups. The maximum absolute flat-plate error is 0.835%. The curvature group motivates an entropy-based empirical within-group correction, while a σf -based empirical within-group correction is considered for the roughness group. Concurrent thickness variation prevents attributing changes in σf solely to roughness. Fitting and assessment on the same specimens describe calibration rather than independent predictive performance. For tapered specimens, increasingly complex spectra accompany an overall reduction in the frequency–design-thickness product. The uniform-plate inversion is compared with design thicknesses, rather than independently measured local thicknesses; a benefit from confidence weighting is not established. The proposed multi-feature-fusion ZGV (MF-ZGV) framework provides a basis for group-specific calibration studies. Reliable automatic classification, independent predictive accuracy, and additive correction under combined conditions remain unverified.