Explainable machine learning for precise fatigue crack tip detection
Abstract Data-driven models based on deep learning have led to tremendous breakthroughs in classical computer vision tasks and have recently made their way into natural sciences. However, the absence of domain knowledge in their inherent design significantly hinders the understanding and acceptance...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2022-06-01
|
| Series: | Scientific Reports |
| Online Access: | https://doi.org/10.1038/s41598-022-13275-1 |
| _version_ | 1818207243524374528 |
|---|---|
| author | David Melching Tobias Strohmann Guillermo Requena Eric Breitbarth |
| author_facet | David Melching Tobias Strohmann Guillermo Requena Eric Breitbarth |
| author_sort | David Melching |
| collection | DOAJ |
| description | Abstract Data-driven models based on deep learning have led to tremendous breakthroughs in classical computer vision tasks and have recently made their way into natural sciences. However, the absence of domain knowledge in their inherent design significantly hinders the understanding and acceptance of these models. Nevertheless, explainability is crucial to justify the use of deep learning tools in safety-relevant applications such as aircraft component design, service and inspection. In this work, we train convolutional neural networks for crack tip detection in fatigue crack growth experiments using full-field displacement data obtained by digital image correlation. For this, we introduce the novel architecture ParallelNets—a network which combines segmentation and regression of the crack tip coordinates—and compare it with a classical U-Net-based architecture. Aiming for explainability, we use the Grad-CAM interpretability method to visualize the neural attention of several models. Attention heatmaps show that ParallelNets is able to focus on physically relevant areas like the crack tip field, which explains its superior performance in terms of accuracy, robustness, and stability. |
| first_indexed | 2024-12-12T04:25:50Z |
| format | Article |
| id | doaj.art-5e3ef24343144fbabbb38ecbe264b687 |
| institution | Directory Open Access Journal |
| issn | 2045-2322 |
| language | English |
| last_indexed | 2024-12-12T04:25:50Z |
| publishDate | 2022-06-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Scientific Reports |
| spelling | doaj.art-5e3ef24343144fbabbb38ecbe264b6872022-12-22T00:38:12ZengNature PortfolioScientific Reports2045-23222022-06-0112111410.1038/s41598-022-13275-1Explainable machine learning for precise fatigue crack tip detectionDavid Melching0Tobias Strohmann1Guillermo Requena2Eric Breitbarth3German Aerospace Center (DLR), Institute of Materials Research, Linder HoeheGerman Aerospace Center (DLR), Institute of Materials Research, Linder HoeheGerman Aerospace Center (DLR), Institute of Materials Research, Linder HoeheGerman Aerospace Center (DLR), Institute of Materials Research, Linder HoeheAbstract Data-driven models based on deep learning have led to tremendous breakthroughs in classical computer vision tasks and have recently made their way into natural sciences. However, the absence of domain knowledge in their inherent design significantly hinders the understanding and acceptance of these models. Nevertheless, explainability is crucial to justify the use of deep learning tools in safety-relevant applications such as aircraft component design, service and inspection. In this work, we train convolutional neural networks for crack tip detection in fatigue crack growth experiments using full-field displacement data obtained by digital image correlation. For this, we introduce the novel architecture ParallelNets—a network which combines segmentation and regression of the crack tip coordinates—and compare it with a classical U-Net-based architecture. Aiming for explainability, we use the Grad-CAM interpretability method to visualize the neural attention of several models. Attention heatmaps show that ParallelNets is able to focus on physically relevant areas like the crack tip field, which explains its superior performance in terms of accuracy, robustness, and stability.https://doi.org/10.1038/s41598-022-13275-1 |
| spellingShingle | David Melching Tobias Strohmann Guillermo Requena Eric Breitbarth Explainable machine learning for precise fatigue crack tip detection Scientific Reports |
| title | Explainable machine learning for precise fatigue crack tip detection |
| title_full | Explainable machine learning for precise fatigue crack tip detection |
| title_fullStr | Explainable machine learning for precise fatigue crack tip detection |
| title_full_unstemmed | Explainable machine learning for precise fatigue crack tip detection |
| title_short | Explainable machine learning for precise fatigue crack tip detection |
| title_sort | explainable machine learning for precise fatigue crack tip detection |
| url | https://doi.org/10.1038/s41598-022-13275-1 |
| work_keys_str_mv | AT davidmelching explainablemachinelearningforprecisefatiguecracktipdetection AT tobiasstrohmann explainablemachinelearningforprecisefatiguecracktipdetection AT guillermorequena explainablemachinelearningforprecisefatiguecracktipdetection AT ericbreitbarth explainablemachinelearningforprecisefatiguecracktipdetection |