TOFD图像时域解混叠的自适应匹配追踪方法*

当TOFD检测的时域分辨率不足时,小缺陷的上下端点衍射波会出现时域混叠,难以准确判断缺陷性质及测量缺陷高度。为解决此问题,提出了自适应匹配追踪算法,该算法利用主成分分析法从检测图像中提取匹配追踪原子,根据散射规律生成字典,通过正交匹配追踪算法实现对时域混叠信号的分离,并使用投影矢量重新成像,实现对缺陷上下端点的有效区分。该方法能够改进TOFD技术在小缺陷测高方面的不足,试验证明,该方法对薄板对接接头中高度为1.80 mm的埋藏缺陷测量误差小于0.30 mm。

Adaptive matching pursuit method for time domain de-aliasing in TOFD image*

It is difficult to determine the type or measure the height of a small defect in a TOFD image with insufficient time resolution, because of the time aliasing between diffraction waves from upper tips and lower tips. To deal with this problem, an adaptive matching pursuit method was proposed. The method comprises the following steps of, abstracting the matching pursuit atom from TOFD image with Principal Component Analysis (PCA); generating the dictionary based on acoustic scattering principle; conducting the time domain de-aliasing by orthogonal matching pursuit; and reconstructing the image with projection coefficients. By the reconstructed image, it is easy to distinguish the upper tip and lower tip of a small defect, which showed a great improvement in small defect inspection than original TOFD technology. A verification experiment was conducted with a thin plate welding specimen, and the measuring error is less than 0.30mm while the height of inner crack is 1.80mm.