| People | Locations | Statistics |
|---|---|---|
| Naji, M. |
| |
| Motta, Antonella |
| |
| Aletan, Dirar |
| |
| Mohamed, Tarek |
| |
| Ertürk, Emre |
| |
| Taccardi, Nicola |
| |
| Kononenko, Denys |
| |
| Petrov, R. H. | Madrid |
|
| Alshaaer, Mazen | Brussels |
|
| Bih, L. |
| |
| Casati, R. |
| |
| Muller, Hermance |
| |
| Kočí, Jan | Prague |
|
| Šuljagić, Marija |
| |
| Kalteremidou, Kalliopi-Artemi | Brussels |
|
| Azam, Siraj |
| |
| Ospanova, Alyiya |
| |
| Blanpain, Bart |
| |
| Ali, M. A. |
| |
| Popa, V. |
| |
| Rančić, M. |
| |
| Ollier, Nadège |
| |
| Azevedo, Nuno Monteiro |
| |
| Landes, Michael |
| |
| Rignanese, Gian-Marco |
|
Mohseni, Ehsan
University of Strathclyde
in Cooperation with on an Cooperation-Score of 37%
Topics
Publications (22/22 displayed)
- 20243-Dimensional residual neural architecture search for ultrasonic defect detectioncitations
- 2023Application of eddy currents for inspection of carbon fibre composites
- 2023Application of machine learning techniques for defect detection, localisation, and sizing in ultrasonic testing of carbon fibre reinforced polymers
- 2023In-process non-destructive evaluation of metal additive manufactured components at build using ultrasound and eddy-current approachescitations
- 2023Mapping SEARCH capabilities to Spirit AeroSystems NDE and automation demand for composites
- 2023Using neural architecture search to discover a convolutional neural network to detect defects From volumetric ultrasonic testing data of composites
- 2023Phased array inspection of narrow-gap weld LOSWF defects for in-process weld inspection
- 2022Transfer learning for classification of experimental ultrasonic non-destructive testing images from synthetic data
- 2022Autonomous and targeted eddy current inspection from UT feature guided wave screening of resistance seam welds
- 2022Mechanical stress measurement using phased array ultrasonic system
- 2022Automated bounding box annotation for NDT ultrasound defect detection
- 2022Multi-sensor electromagnetic inspection feasibility for aerospace composites surface defects
- 2022Investigating ultrasound wave propagation through the coupling medium and non-flat surface of wire + arc additive manufactured components inspected by a PAUT roller-probe
- 2022Automated multi-modal in-process non-destructive evaluation of wire + arc additive manufacturing
- 2022Dual-tandem phased array inspection for imaging near-vertical defects in narrow gap welds
- 2022Targeted eddy current inspection based on ultrasonic feature guided wave screening of resistance seam welds
- 2022In-process non-destructive evaluation of wire + arc additive manufacture components using ultrasound high-temperature dry-coupled roller-probe
- 2022Collaborative robotic Wire + Arc Additive Manufacture and sensor-enabled in-process ultrasonic Non-Destructive Evaluationcitations
- 2022Automated real time eddy current array inspection of nuclear assetscitations
- 2020In-process calibration of a non-destructive testing system used for in-process inspection of multi-pass weldingcitations
- 2020Laser-assisted surface adaptive ultrasound (SAUL) inspection of samples with complex surface profiles using a phased array roller-probe
- 2019Ultrasonic phased array inspection of a Wire + Arc Additive Manufactured (WAAM) sample with intentionally embedded defectscitations
Places of action
| Organizations | Location | People |
|---|
article
3-Dimensional residual neural architecture search for ultrasonic defect detection
Abstract
This study presents a deep learning methodology using 3-dimensional (3D) convolutional neural networks to detect defects in carbon fiber reinforced polymer composites through volumetric ultrasonic testing data. Acquiring large amounts of ultrasonic training data experimentally is expensive and time-consuming. To address this issue, a synthetic data generation method was extended to incorporate volumetric data. By preserving the complete volumetric data, complex preprocessing is reduced, and the model can utilize spatial and temporal information that is lost during imaging. This enables the model to utilize important features that might be overlooked otherwise.<br/><br/>The performance of three architectures were compared. The first architecture is prevalent in the literature for the classification of volumetric datasets. The second demonstrated a hand-designed approach to architecture design, with modifications to the first architecture to address the challenges of this specific task. A key modification was the use of cuboidal kernels to account for the large aspect ratios seen in ultrasonic data. The third architecture was discovered through neural architecture search from a modified 3D Residual Neural Network (ResNet) search space. Additionally, domain-specific augmentation methods were incorporated during training, resulting in significant improvements in model performance, with a mean accuracy improvement of 22.4% on the discovered architecture. The discovered architecture demonstrated the best performance with a mean accuracy increase of 7.9% over the second best model. It was able to consistently detect all defects whilst maintaining a model size smaller than most 2-dimensional (2D) ResNets. Each model had an inference time of less than 0.5 seconds, making them efficient for the interpretation of large amounts of data.<br/>