A review of synthetic and augmented training data for machine learning in ultrasonic non-destructive evaluation

Uhlig, Sebastian; Alkhasli, Ilkin; Schubert, Frank; Tschöpe, Constanze; Wolff, Matthias

doi:10.1016/j.ultras.2023.107041

Cited by 13 publications

(2 citation statements)

References 155 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Ultrasonic testing is a technique that employs ultrasonic waves to detect macroscopic defects, characterize changes in material structure and mechanical properties, and measure the geometric features of a workpiece [38]. This method boasts advantages such as high detection speed, precision, directivity, sound wave energy, and strong penetration capability [39].…”

Section: Ultrasonic Testingmentioning

confidence: 99%

Systematic Evaluation of Ultrasonic In-Line Inspection Techniques for Oil and Gas Pipeline Defects Based on Bibliometric Analysis

Huang,

Chen,

et al. 2024

Sensors

View full text Add to dashboard Cite

The global reliance on oil and gas pipelines for energy transportation is increasing. As the pioneering review in the field of ultrasonic defect detection for oil and gas pipelines based on bibliometric methods, this study employs visual analysis to identify the most influential countries, academic institutions, and journals in this domain. Through cluster analysis, it determines the primary trends, research hotspots, and future directions in this critical field. Starting from the current global industrial ultrasonic in-line inspection (ILI) detection level, this paper provides a flowchart for selecting detection methods and a table for defect comparison, detailing the comparative performance limits of different detection devices. It offers a comprehensive perspective on the latest ultrasonic pipeline detection technology from laboratory experiments to industrial practice.

show abstract

Section: Ultrasonic Testingmentioning

confidence: 99%

Systematic Evaluation of Ultrasonic In-Line Inspection Techniques for Oil and Gas Pipeline Defects Based on Bibliometric Analysis

Huang,

Chen,

et al. 2024

Sensors

View full text Add to dashboard Cite

show abstract

“…Advanced post-processing algorithms of the collected data have been developed to provide C-scan images of the interface defects in multilayer structures, improve detection quality or extract features from the signal responses [ 2 , 3 , 16 , 19 , 20 , 23 ]. Furthermore, machine learning is also being applied to study the defects in the structures [ 1 , 24 , 25 , 26 ]. In our previous works [ 2 ], a novel post-processing technique was developed for the detection of disbonds in multilayered structures and it eliminated some influential factors and improved detectability.…”

Section: Introductionmentioning

confidence: 99%

Ultrasonic Features for Evaluation of Adhesive Joints: A Comparative Study of Interface Defects

Smagulova,

Yilmaz,

Jasiuniene

2023

Sensors

View full text Add to dashboard Cite

Ultrasonic non-destructive evaluation in pulse-echo mode is used for the inspection of single-lap aluminum adhesive joints, which contain interface defects in bonding area. The aim of the research is to increase the probability of defect detection in addition to ensuring that the defect sizes are accurately estimated. To achieve this, this study explores additional ultrasonic features (not only amplitude) that could provide more accurate information about the quality of the structure and the presence of interface defects. In this work, two types of interface defects, namely inclusions and delaminations, were studied based on the extracted ultrasonic features in order to evaluate the expected feasibility of defect detection and the evaluation of its performance. In addition, an analysis of multiple interface reflections, which have been proved to improve detection in our previous works, was applied along with the extraction of various ultrasonic features, since it can increase the probability of defect detection. The ultrasonic features with the best performance for each defect type were identified and a comparative analysis was carried out, showing that it is more challenging to size inclusion-type defects compared to delaminations. The best performance is observed for the features such as peak-to-peak amplitude, ratio coefficients, absolute energy, absolute time of flight, mean value of the amplitude, standard deviation value, and variation coefficient for both types of defects. The maximum relative error of the defect size compared to the real one for these features is 16.9% for inclusions and 3.6% for delaminations, with minimum errors of 11.4% and 2.2%, respectively. In addition, it was determined that analysis of the data from repetitive reflections from the sample interface, namely, the aluminum-adhesive second and third reflections, that these contribute to an increase in the probability of defect detection.

show abstract

Automatic PAUT Crack Detection and Depth Identification Framework Based on Inspection Robot and Deep Learning Method

Hu,

Gou,

Yang

et al. 2024

Journal of Infrastructure Intelligence and Resilience

View full text Add to dashboard Cite

A review of synthetic and augmented training data for machine learning in ultrasonic non-destructive evaluation

Cited by 13 publications

References 155 publications

Systematic Evaluation of Ultrasonic In-Line Inspection Techniques for Oil and Gas Pipeline Defects Based on Bibliometric Analysis

Systematic Evaluation of Ultrasonic In-Line Inspection Techniques for Oil and Gas Pipeline Defects Based on Bibliometric Analysis

Ultrasonic Features for Evaluation of Adhesive Joints: A Comparative Study of Interface Defects

Automatic PAUT Crack Detection and Depth Identification Framework Based on Inspection Robot and Deep Learning Method

Contact Info

Product

Resources

About