Augmented Ultrasonic Data for Machine Learning
Аугментированные ультразвуковые данные для машинного обучения
2021-01-02
SCID: 54.1/5vddunm9
Discuss with AI
deep convolutional networksflaw detectionnon-destructive testingphased-array ultrasonic dataultrasonic data augmentation
Figures from the paper
Abstract (AI)
Abstract Flaw detection in non-destructive testing, especially for complex signals like ultrasonic data, has thus far relied heavily on the expertise and judgement of trained human inspectors. While automated systems have been used for a long time, these have mostly been limited to using simple decision automation, such as signal amplitude threshold. The recent advances in various machine learning algorithms have solved many similarly difficult classification problems, that have previously been considered intractable. For non-destructive testing, encouraging results have already been reported in the open literature, but the use of machine learning is still very limited in NDT applications in the field. Key issue hindering their use, is the limited availability of representative flawed data-sets to be used for training. In the present paper, we develop modern, deep convolutional network to detect flaws from phased-array ultrasonic data. We make extensive use of data augmentation to enhance the initially limited raw data and to aid learning. The data augmentation utilizes virtual flaws—a technique, that has successfully been used in training human inspectors and is soon to be used in nuclear inspection qualification. The results from the machine learning classifier are compared to human performance. We show, that using sophisticated data augmentation, modern deep learning networks can be trained to achieve human-level performance.
Key Findings
1
A modern deep convolutional network was developed to detect flaws in phased-array ultrasonic data.
2
Extensive data augmentation using virtual flaws addressed the limited availability of representative flawed datasets for training.
3
Sophisticated augmentation enabled the deep-learning classifier to achieve performance comparable to human inspectors.
4
The study compares automated flaw-detection results directly with human performance in non-destructive testing.
Research Object
Flaw detection in phased-array ultrasonic data for non-destructive testing
Research Subject
human-level flaw-detection performance of deep convolutional networks enabled by data augmentation with virtual flaws
Publication Details
Publication Date
2021-01-02
Journal
Publisher
ISSN
Cited by
195
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest
References available in scid.ai2
Cited by5
NDE 4.0—A Design Thinking Perspective2021
Automated defect detection in digital radiography of aerospace welds using deep learning2022
Application of Object Detection Algorithms in Non-Destructive Testing of Pressure Equipment: A Review2024
Leveraging Segment Anything Model (SAM) for Weld Defect Detection in Industrial Ultrasonic B-Scan Images2025
Advanced Non-Destructive Testing Simulation and Modeling Approaches for Fiber-Reinforced Polymer Pipes: A Review2025