Salvato in:
| Autori principali: | , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2604.11494 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866911588850597888 |
|---|---|
| author | Yang, Zhangke Meng, Zhaoxu |
| author_facet | Yang, Zhangke Meng, Zhaoxu |
| contents | Tendon-bone enthesis connects tendon and bone, two mechanically dissimilar materials, while effectively minimizing stress concentrations, a capability rarely achieved in engineering materials. Its hierarchical organization and graded variations in composition or mineralization are widely recognized as key contributors to its exceptional performance. Here, we investigate the mechanics of enthesis, focusing on the insertion of interface collagen fibers into bone where hierarchical collagen fibril structures and graded mineralization are present, and translate these insights into bioinspired engineering material design using a convolutional neural network-based field predictor (CNNFP). We first construct a three-dimensional finite element model (FEM) of the interface fiber-bone enthesis, in which local material properties depend on mineralization level, mean fibril orientation, and angular dispersion, informed by a multiscale continuum theory. We introduce a scalar risk factor that integrates local stress states and constituent fibril organizations to quantify local vulnerability. Simulation results demonstrate that graded and spatially heterogeneous configurations markedly reduce stress concentrations, supporting prevailing biomechanical hypotheses. We then train the CNNFP as an accurate surrogate for FEM and embed it within a kernel-based gradient optimization framework to efficiently identify optimal field configurations. The optimized designs are validated against FEM ground truth, establishing a generalizable AI-enabled pathway for the optimization of bioinspired functionally graded materials. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_11494 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Machine Learning-Enabled Mechanical Analysis and Optimization of Bioinspired Functionally Graded Materials Yang, Zhangke Meng, Zhaoxu Soft Condensed Matter Tendon-bone enthesis connects tendon and bone, two mechanically dissimilar materials, while effectively minimizing stress concentrations, a capability rarely achieved in engineering materials. Its hierarchical organization and graded variations in composition or mineralization are widely recognized as key contributors to its exceptional performance. Here, we investigate the mechanics of enthesis, focusing on the insertion of interface collagen fibers into bone where hierarchical collagen fibril structures and graded mineralization are present, and translate these insights into bioinspired engineering material design using a convolutional neural network-based field predictor (CNNFP). We first construct a three-dimensional finite element model (FEM) of the interface fiber-bone enthesis, in which local material properties depend on mineralization level, mean fibril orientation, and angular dispersion, informed by a multiscale continuum theory. We introduce a scalar risk factor that integrates local stress states and constituent fibril organizations to quantify local vulnerability. Simulation results demonstrate that graded and spatially heterogeneous configurations markedly reduce stress concentrations, supporting prevailing biomechanical hypotheses. We then train the CNNFP as an accurate surrogate for FEM and embed it within a kernel-based gradient optimization framework to efficiently identify optimal field configurations. The optimized designs are validated against FEM ground truth, establishing a generalizable AI-enabled pathway for the optimization of bioinspired functionally graded materials. |
| title | Machine Learning-Enabled Mechanical Analysis and Optimization of Bioinspired Functionally Graded Materials |
| topic | Soft Condensed Matter |
| url | https://arxiv.org/abs/2604.11494 |