A novel characteristic slope-based defect depth inversion method for pipelines using the flexible honeycomb-shaped eddy current sensor

Eddy current testing is critical for industrial pipeline safety. The measurement of defect depth on curved surfaces is influenced by both defect orientations and lengths. In this paper, a novel characteristic slope-based defect depth estimation method is proposed using the designed flexible honeycomb-shaped sensor. The designed flexible honeycomb-shaped sensor with the adjacent anti-phase excitation strategy can enhance magnetic field strength and suppress lift-off noise. From the detected signals, the signal trajectories in the complex plane exhibit a slope invariance with respect to defect orientations and lengths, while maintaining a distinct monotonic dependency on defect depth. Based on this slope feature, a defect depth inversion model is constructed employing the Natural Gradient Boosting algorithm. Experiments have been conducted on aluminium alloy pipes containing defects of varying lengths and orientations. The results demonstrate that the method effectively overcomes the interference of defect orientations and lengths. It achieves high precision depth assessment with an average prediction accuracy reaching 98.8%.