Speaker
Description
Active infrared thermography is widely adopted for non-destructive evaluation of composites, yet faces challenges including thermal diffusion-induced defect boundary blurring, excitation signal modulation effects on signal-to-noise ratio (SNR), and limited 3D defect morphology reconstruction accuracy. To address these limitations, an adaptive tomographic radar thermography (ATRT) method is proposed. ATRT introduces pixel-level membership to distinctly separate defect and healthy areas, achieving precise lateral boundary definition. The large time-bandwidth product pulse radar modulation is employed to ensure high-SNR excitation, while surface temperature differential-depth correlations is established for accurate depth inversion and complete 3D defect reconstruction. Frist, the defect detection process is formulated as a nonlinear optimization problem, solved through the integration of genetic and simulated annealing algorithms. Secondly, the relationship between defect depth and maximum temperature difference is derived from thermal conduction principles. Next, two complementary probability of detection models ATPE-POD and ERE-POD are formulated using asymptotic theory and resampling techniques, respectively, to evaluate diameter-to-depth ratio effects on detection capability. Finally, experimental validation on composite laminates reveals ATRT's capability. Results demonstrate that ATRT shows 90% detection probability for defects with diameter-to-depth ratios exceeding 1.11, while achieving 2-4 times higher SNR than traditional methods. The technique maintains under 10% radial size error in 3D reconstruction. Depth reconstruction accuracy exhibits a quasi-linear dependence on defect lateral dimensions, showing error reduction from 20% at 1.25 mm diameter to 8% at 7.0 mm diameter, while maintaining submillimeter depth resolution. ATRT provides an effective solution for structural health monitoring of composite materials, significantly advancing defect characterization capabilities.
