Mechanical dissimilarity of defects in welded joints via Grassmann manifold and machine learning
Assessing the harmfulness of defects based on images is becoming more and more common in industry. At present, these defects can be inserted in digital twins that aim to replicate in a mechanical model what is observed on a component so that an image-based diagnosis can be further conducted. However...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Académie des sciences
2020-11-01
|
| Series: | Comptes Rendus. Mécanique |
| Subjects: | |
| Online Access: | https://comptes-rendus.academie-sciences.fr/mecanique/articles/10.5802/crmeca.51/ |
| _version_ | 1797651440823435264 |
|---|---|
| author | Ryckelynck, David Goessel, Thibault Nguyen, Franck |
| author_facet | Ryckelynck, David Goessel, Thibault Nguyen, Franck |
| author_sort | Ryckelynck, David |
| collection | DOAJ |
| description | Assessing the harmfulness of defects based on images is becoming more and more common in industry. At present, these defects can be inserted in digital twins that aim to replicate in a mechanical model what is observed on a component so that an image-based diagnosis can be further conducted. However, the variety of defects, the complexity of their shape, and the computational complexity of finite element models related to their digital twin make this kind of diagnosis too slow for any practical application. We show that a classification of observed defects enables the definition of a dictionary of digital twins. These digital twins prove to be representative of model-reduction purposes while preserving an acceptable accuracy for stress prediction. Nonsupervised machine learning is used for both the classification issue and the construction of reduced digital twins. The dictionary items are medoids found by a k-medoids clustering algorithm. Medoids are assumed to be well distributed in the training dataset according to a metric or a dissimilarity measurement. In this paper, we propose a new dissimilarity measurement between defects. It is theoretically founded according to approximation errors in hyper-reduced predictions. In doing so, defect classes are defined according to their mechanical effect and not directly according to their morphology. In practice, each defect in the training dataset is encoded as a point on a Grassmann manifold. This methodology is evaluated through a test set of observed defects totally different from the training dataset of defects used to compute the dictionary of digital twins. The most appropriate item in the dictionary for model reduction is selected according to an error indicator related to the hyper-reduced prediction of stresses. No plasticity effect is considered here (merely isotropic elastic materials), which is a strong assumption but which is not critical for the purpose of this work. In spite of the large variety of defects, we provide accurate predictions of stresses for most of defects in the test set. |
| first_indexed | 2024-03-11T16:14:49Z |
| format | Article |
| id | doaj.art-053b038bb6604035996b83e9c852fe0a |
| institution | Directory Open Access Journal |
| issn | 1873-7234 |
| language | English |
| last_indexed | 2024-03-11T16:14:49Z |
| publishDate | 2020-11-01 |
| publisher | Académie des sciences |
| record_format | Article |
| series | Comptes Rendus. Mécanique |
| spelling | doaj.art-053b038bb6604035996b83e9c852fe0a2023-10-24T14:21:01ZengAcadémie des sciencesComptes Rendus. Mécanique1873-72342020-11-0134810-1191193510.5802/crmeca.5110.5802/crmeca.51Mechanical dissimilarity of defects in welded joints via Grassmann manifold and machine learningRyckelynck, David0https://orcid.org/0000-0003-3268-4892Goessel, Thibault1Nguyen, Franck2Mines ParisTech PSL University, Centre des Matériaux, Evry, FranceMines ParisTech PSL University, FranceMines ParisTech PSL University, Centre des Matériaux, Evry, FranceAssessing the harmfulness of defects based on images is becoming more and more common in industry. At present, these defects can be inserted in digital twins that aim to replicate in a mechanical model what is observed on a component so that an image-based diagnosis can be further conducted. However, the variety of defects, the complexity of their shape, and the computational complexity of finite element models related to their digital twin make this kind of diagnosis too slow for any practical application. We show that a classification of observed defects enables the definition of a dictionary of digital twins. These digital twins prove to be representative of model-reduction purposes while preserving an acceptable accuracy for stress prediction. Nonsupervised machine learning is used for both the classification issue and the construction of reduced digital twins. The dictionary items are medoids found by a k-medoids clustering algorithm. Medoids are assumed to be well distributed in the training dataset according to a metric or a dissimilarity measurement. In this paper, we propose a new dissimilarity measurement between defects. It is theoretically founded according to approximation errors in hyper-reduced predictions. In doing so, defect classes are defined according to their mechanical effect and not directly according to their morphology. In practice, each defect in the training dataset is encoded as a point on a Grassmann manifold. This methodology is evaluated through a test set of observed defects totally different from the training dataset of defects used to compute the dictionary of digital twins. The most appropriate item in the dictionary for model reduction is selected according to an error indicator related to the hyper-reduced prediction of stresses. No plasticity effect is considered here (merely isotropic elastic materials), which is a strong assumption but which is not critical for the purpose of this work. In spite of the large variety of defects, we provide accurate predictions of stresses for most of defects in the test set.https://comptes-rendus.academie-sciences.fr/mecanique/articles/10.5802/crmeca.51/Data encodingHyper-reductionReduced order modelROM-netTaxonomy of defectsComputer vision |
| spellingShingle | Ryckelynck, David Goessel, Thibault Nguyen, Franck Mechanical dissimilarity of defects in welded joints via Grassmann manifold and machine learning Comptes Rendus. Mécanique Data encoding Hyper-reduction Reduced order model ROM-net Taxonomy of defects Computer vision |
| title | Mechanical dissimilarity of defects in welded joints via Grassmann manifold and machine learning |
| title_full | Mechanical dissimilarity of defects in welded joints via Grassmann manifold and machine learning |
| title_fullStr | Mechanical dissimilarity of defects in welded joints via Grassmann manifold and machine learning |
| title_full_unstemmed | Mechanical dissimilarity of defects in welded joints via Grassmann manifold and machine learning |
| title_short | Mechanical dissimilarity of defects in welded joints via Grassmann manifold and machine learning |
| title_sort | mechanical dissimilarity of defects in welded joints via grassmann manifold and machine learning |
| topic | Data encoding Hyper-reduction Reduced order model ROM-net Taxonomy of defects Computer vision |
| url | https://comptes-rendus.academie-sciences.fr/mecanique/articles/10.5802/crmeca.51/ |
| work_keys_str_mv | AT ryckelynckdavid mechanicaldissimilarityofdefectsinweldedjointsviagrassmannmanifoldandmachinelearning AT goesselthibault mechanicaldissimilarityofdefectsinweldedjointsviagrassmannmanifoldandmachinelearning AT nguyenfranck mechanicaldissimilarityofdefectsinweldedjointsviagrassmannmanifoldandmachinelearning |