期刊
IEEE TRANSACTIONS ON ULTRASONICS FERROELECTRICS AND FREQUENCY CONTROL
卷 68, 期 5, 页码 1854-1865出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TUFFC.2020.3045847
关键词
Inspection; Deep learning; Acoustics; Training; Surface cracks; Complexity theory; Arrays; Deep learning; defect characterization; neural networks; plane wave imaging (PWI); simulation; ultrasound
资金
- Engineering and Physical Sciences Research Council (EPSRC) via the Research Centre for Non-Destructive Evaluation (RCNDE) [EP/L015587/1]
- Baker Hughes, Cramlington, U.K.
Machine learning has the potential to significantly improve defect characterization accuracy in nondestructive evaluation, but scarce real defect data has hindered its application in NDE. This article demonstrates the use of a hybrid simulation method to train a neural network for characterizing real defects, showing that deep learning outperforms traditional image-based sizing methods in accuracy for crack characterization.
Machine learning for nondestructive evaluation (NDE) has the potential to bring significant improvements in defect characterization accuracy due to its effectiveness in pattern recognition problems. However, the application of modern machine learning methods to NDE has been obstructed by the scarcity of real defect data to train on. This article demonstrates how an efficient, hybrid finite element (FE) and ray-based simulation can be used to train a convolutional neural network (CNN) to characterize real defects. To demonstrate this methodology, an inline pipe inspection application is considered. This uses four plane wave images from two arrays and is applied to the characterization of cracks of length 1-5 mm and inclined at angles of up to 20 degrees from the vertical. A standard image-based sizing technique, the 6-dB drop method, is used as a comparison point. For the 6-dB drop method, the average absolute error in length and angle prediction is +/- 1.1 mm and +/- 8.6 degrees, respectively, while the CNN is almost four times more accurate at +/- 0.29 mm and +/- 2.9 degrees. To demonstrate the adaptability of the deep learning approach, an error in sound speed estimation is included in the training and test set. With a maximum error of 10% in shear and longitudinal sound speed, the 6-dB drop method has an average error of +/- 1.5 mmm and +/- 12 degrees, while the CNN has +/- 0.45 mm and +/- 3.0 degrees. This demonstrates far superior crack characterization accuracy by using deep learning rather than traditional image-based sizing.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据