A deep neural network physics-based reduced order model for dynamic stall
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866915779290595328 |
|---|---|
| author | Baldan, Giacomo Guardone, Alberto |
| author_facet | Baldan, Giacomo Guardone, Alberto |
| contents | A physics-based machine learning framework is developed to compute the aerodynamic forces and moment for a pitching NACA0012 airfoil incurring in light and deep dynamic stall. Three deep neural network frameworks of increasing complexity are investigated: two multilayer perceptrons and a convolutional neural network. The convolutional framework, in addition to the standard mean squared error loss, features a physically-informed improved loss function to compute the airfoil loads. In total, four models are investigated of increasingly complexity. The convolutional model, coupled with the physics-based loss function, is found to robustly and efficiently predict pressure and skin friction distributions over the airfoil over the entire pitching cycle. Periodic conditions are implemented to grant the physical smoothness of the model output both in space and time. An analysis of the training dataset point distributions is performed to point out the effects of adopting low discrepancy sequences, such as Latin hypercube, Sobol', and Halton, compared to random and uniform sequences. The current model shows unprecedented performances in predicting forces and pitching moment in a broad range of operating conditions. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2401_14728 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | A deep neural network physics-based reduced order model for dynamic stall Baldan, Giacomo Guardone, Alberto Fluid Dynamics A physics-based machine learning framework is developed to compute the aerodynamic forces and moment for a pitching NACA0012 airfoil incurring in light and deep dynamic stall. Three deep neural network frameworks of increasing complexity are investigated: two multilayer perceptrons and a convolutional neural network. The convolutional framework, in addition to the standard mean squared error loss, features a physically-informed improved loss function to compute the airfoil loads. In total, four models are investigated of increasingly complexity. The convolutional model, coupled with the physics-based loss function, is found to robustly and efficiently predict pressure and skin friction distributions over the airfoil over the entire pitching cycle. Periodic conditions are implemented to grant the physical smoothness of the model output both in space and time. An analysis of the training dataset point distributions is performed to point out the effects of adopting low discrepancy sequences, such as Latin hypercube, Sobol', and Halton, compared to random and uniform sequences. The current model shows unprecedented performances in predicting forces and pitching moment in a broad range of operating conditions. |
| title | A deep neural network physics-based reduced order model for dynamic stall |
| topic | Fluid Dynamics |
| url | https://arxiv.org/abs/2401.14728 |