Deep Learning for Ultrasonic Crack Characterization in NDE

TLDR

Machine learning can greatly improve defect characterization in nondestructive evaluation, but progress has been limited by the scarcity of real defect data. This study shows that a hybrid finite‑element and ray‑based simulation can train a convolutional neural network to accurately characterize real pipe cracks, and that the method remains robust when sound‑speed errors are introduced. The authors generated synthetic pipe‑inspection data with four plane‑wave images from two arrays, trained a CNN to predict crack length (1–5 mm) and inclination (up to 20°), and compared its performance to the 6‑dB drop sizing technique while also testing robustness to up to 10 % sound‑speed errors. The CNN outperformed the 6‑dB drop method, reducing length error to ±0.29 mm and angle error to ±2.9°, and maintained superior accuracy (±0.45 mm, ±3.0°) even when sound‑speed errors of 10 % were present.

Abstract

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° 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°, respectively, while the CNN is almost four times more accurate at ±0.29 mm and ±2.9°. 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°, while the CNN has ±0.45 mm and ±3.0°. This demonstrates far superior crack characterization accuracy by using deep learning rather than traditional image-based sizing.