Recognition of weld damage in T-welded steel based on a modified reconstruction algorithm for probabilistic inspection of damage

Using ultrasonic guided wave(UGW) technology to realize on-line crack damage identification of T-welded steel has been a research hotspot in recent years. However, in practical engineering applications, the localization results of the traditional reconstruction algorithm for probabilistic inspection of damage (RAPID) are affected by the path sparsity. In addition, the modes of the guided wave (GW) signals are complex and seriously aliased, which makes it challenging to apply high-precision damage localization methods based on time-of-flight (TOF), such as the time-reversal method. To this end, this paper proposes a modified RAPID based on imaging path pairs. A T-welded steel experiment system is built, and firstly, the laws of GW signals under different excitation parameters are analyzed. After that, the effects of the welds on various types of path signals are determined. Then, the subregion where the damage is located is determined based on the damage indices (DIs) of the constituent paths and assessment paths. Finally, probabilistic imaging is performed using imaging path pairs. Simulated and real damages are utilized to validate the effectiveness of the proposed method. The results show that the average localization error of the proposed method is 5.8 mm, which is a significant advantage over the traditional RAPID and TOF-based RAPID.