Quantitative mapping of depth profile of fatigue cracks using eddy current pulsed thermography assisted by PCA and 2D wavelet transformation

Abstract

Fatigue cracks are widely generated in numerous engineering components under alternating loads, whose sudden brittle fracture would seriously shorten the service life and cause significant hazards. A large number of nondestructive testing (NDT) techniques have been utilized to quantitatively evaluate the fatigue cracks, however, most of these methods are difficult to be used for a full characterization of the cracks’ depth profile. The eddy current pulsed thermography (ECPT) as a promising NDT technique shows great potential for fatigue crack evaluation due to its high detection efficiency, large inspection area, and high spatial resolution. In this article, a ferrite yoke was utilized to generate a homogeneous eddy current field, and as such to enhance the detection sensitivity and to provide a wide open-view area for the camera. The hybrid processing strategy combining principal component analysis (PCA) and 2D wavelet transformation was proposed to extract spatial features of the thermographic image sequences. Firstly, numerical results demonstrate that the profiles of slot cracks can be identified directly from the thermographic images in the early heating stage. Afterward, the ECPT experiments on slot cracks with different depth profiles demonstrate the feasibility of the hybrid processing strategy. Then, three fatigue cracks were manufactured through the three-point bending method and detected by ECPT. The results of the hybrid processing method indicate that these three fatigue cracks have elliptical profiles. Finally, the authentic depth profile of a fatigue crack with an elliptical shape is extracted through a destructive experiment. The comparison results between the real depth profile and the profile characterized through the proposed strategy show great consistency and validate the superiority of the proposed strategy.