A graphic augmented defect recognition algorithm was developed for phased array ultrasonic testing (PAUT) on tubular TKY joints. Based on PAUT sector scan (S-scan), 2D maps of the scans are obtained in terms of signal intensity and locations using spatial clustering and segmentation. The transformed S-scan data was further mapped with weld geometrical model to discriminate defect indications from the signals due to weld boundaries. Geometrical features of the identified defect indications were extracted to estimate the size and depth of the defects. The algorithm was verified by experiments on a TKY artefact with purposely-made defects. It shows that our algorithm was able to identify and locate all the defects in the artefact. Compared with the results obtained manually using the standard ultrasonic testing method, the differences of our results in terms of the defect size and depth were within 2 mm. The algorithm, coded into a software tool, demonstrated that the defects in the tubular TKY welds can be detected automatically and visualized in an intuitive manner.

中文翻译：

---

管状 TKY 接头相控阵超声检测的图形增强缺陷识别

为管状 TKY 接头上的相控阵超声检测 (PAUT) 开发了一种图形增强缺陷识别算法。基于 PAUT 扇区扫描 (S-scan)，使用空间聚类和分割根据信号强度和位置获得扫描的 2D 地图。转换后的 S 扫描数据与焊缝几何模型进一步映射，以区分由于焊缝边界引起的信号中的缺陷指示。提取已识别缺陷迹象的几何特征以估计缺陷的大小和深度。该算法通过对带有故意制造缺陷的 TKY 人工制品的实验进行了验证。它表明我们的算法能够识别和定位人工制品中的所有缺陷。与使用标准超声波检测方法手动获得的结果相比，我们的结果在缺陷尺寸和深度方面的差异在 2 毫米以内。该算法编码到软件工具中，证明可以自动检测管状 TKY 焊缝中的缺陷，并以直观的方式可视化。