Diffraction time difference method (TOFD) has the advantages of rich detection information, strong noise resistance, high efficiency and accurate positioning, and is widely used in weld detection, while TOFD map evaluation mainly depends on people, there are low efficiency, subjectivity, low reliability and other problems. In order to improve the accuracy and efficiency of defect type recognition, combined with TOFD map and waveform characteristics, a cascading fusion converge neural network (CNN-TCN) is constructed to automatically identify defect types in weld defect scanning images. The results show that the classification accuracy of CNN-TCN is significantly higher than that of the simple lenet5 and TCN methods, and the recognition rate of defect types can reach more than 88%, which shows that the image characteristics of TOFD weld defects and the waveform characteristics of their ultrasonic sequence are considered comprehensively, which has a great effect on improving the classification accuracy of TOFD weld defects, and has high recognition rate, robustness and anti-jamming ability.
正在翻译中..