Temporal-spatial enhanced eddy current pulsed thermography for detecting defects in CFRP-reinforced steel structures

Carbon fiber reinforced polymer (CFRP)-reinforced steel structures (CRSS) play a vital role in strengthening and rehabilitating civil infrastructure. While eddy current pulsed thermography (ECPT) is applicable for inspecting CRSS, it often fails to detect defects with weak thermal indications. To address this limitation, a novel temporal-spatial enhanced ECPT (TS-ECPT) framework is proposed for accurate non-destructive quality assessment of CRSS. The TS-ECPT framework implements a progressive enhancement strategy. Specifically, a physics-driven dual-current excitation strategy is first developed to generate precisely aligned training pairs, capturing the transition from weak to strong defect indications. Subsequently, Initial Baseline Subtraction (IBS) is applied in the temporal domain to physically decouple specimen-specific interference and extract defect features, providing an initial enhancement and critical signals for the subsequent stage. Building upon this, a deep learning-based mapping model further refines and enhances these features in the spatial domain to achieve precise defect indications. Experimental results validate that the proposed synergistic enhancement strategy within TS-ECPT framework significantly improves detection sensitivity, with a signal-to-noise ratio enhancement of 64.2–414.6 % relative to the best-performing benchmark method. This work establishes a reliable and high-sensitivity solution for quality assessment of CRSS, contributing to the safety and service life of these structures. Future research will focus on developing intelligent, automated inspection systems for the quantitative assessment of in-service CRSS.