Circumferential Shear Horizontal Guided Wave Crack Inspection of 3PE-Coated Pipes Based on a CNN.

Circumferential shear horizontal (CSH) guided waves propagate fully parallel to the surface of a pipe and are thus ideal for axial crack detection. At present, systems using EMATs as transmitters and receivers for in-line inspection (ILI) are widely available. Previous circumferential wave inspection methods led to the development of the bare pipe model and considered the impact of a thin coating on attenuation. This paper focuses on the influence of a three-layer polyethylene (3PE) coating on CSH guided wave inspection. A comparison of the guided wave motions between the multilayer hollow cylinder model of a 3PE-coated pipe and the bare pipe model reveals an increase in the number of admissible modes. Experiments were performed on bare and 3PE-coated pipes with axial defects at two typical excitation frequencies that may be feasible for detection. The results show that the defect echo signal is easier to distinguish in the time domain at low frequencies; however, based on the wavelet time–frequency spectrum and a convolutional neural network (CNN), an axial defect in a 3PE-coated pipe can be identified more effectively at a higher frequency. This work considering the effects of a thick coating achieves high-resolution detection and provides insight into the practical automatic inspection of oil and gas pipelines by means of deep learning techniques. [ABSTRACT FROM AUTHOR]