Damage detection near the plate boundary based on mode separation algorithm of Lamb waves

Lamb waves are effective for damage detection in large plate-like structures. To prevent boundary reflections from contaminating the damage reflection near the plate edge, a mode separation algorithm is proposed. By predicting and removing the theoretical waveforms of boundary reflections, the damage reflection near the plate edge can be accurately extracted. In addition, the Isolation Forest algorithm is introduced to facilitate defect identification by suppressing false pixels and sharpening defect features. Both numerical simulations and experiments are implemented to validate the proposed method. Damage near the plate edge is successfully detected with an absolute error of less than 4 mm. The proposed method can eliminate the blind zone of ultrasonic testing near the structural edge and can improve the quality of the damage image for plate-like structures.Footnote11. delete the first email address of the corresponding author