Strategies for multi-case physics-informed neural networks for tube flows: a study using 2D flow scenarios
Abstract Fluid dynamics computations for tube-like geometries are crucial in biomedical evaluations of vascular and airways fluid dynamics. Physics-Informed Neural Networks (PINNs) have emerged as a promising alternative to traditional computational fluid dynamics (CFD) methods. However, vanilla PIN...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2024-05-01
|
| Series: | Scientific Reports |
| Subjects: | |
| Online Access: | https://doi.org/10.1038/s41598-024-62117-9 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1846101297096294400 |
|---|---|
| author | Hong Shen Wong Wei Xuan Chan Bing Huan Li Choon Hwai Yap |
| author_facet | Hong Shen Wong Wei Xuan Chan Bing Huan Li Choon Hwai Yap |
| author_sort | Hong Shen Wong |
| collection | DOAJ |
| description | Abstract Fluid dynamics computations for tube-like geometries are crucial in biomedical evaluations of vascular and airways fluid dynamics. Physics-Informed Neural Networks (PINNs) have emerged as a promising alternative to traditional computational fluid dynamics (CFD) methods. However, vanilla PINNs often demand longer training times than conventional CFD methods for each specific flow scenario, limiting their widespread use. To address this, multi-case PINN approach has been proposed, where varied geometry cases are parameterized and pre-trained on the PINN. This allows for quick generation of flow results in unseen geometries. In this study, we compare three network architectures to optimize the multi-case PINN through experiments on a series of idealized 2D stenotic tube flows. The evaluated architectures include the ‘Mixed Network’, treating case parameters as additional dimensions in the vanilla PINN architecture; the “Hypernetwork”, incorporating case parameters into a side network that computes weights in the main PINN network; and the “Modes” network, where case parameters input into a side network contribute to the final output via an inner product, similar to DeepONet. Results confirm the viability of the multi-case parametric PINN approach, with the Modes network exhibiting superior performance in terms of accuracy, convergence efficiency, and computational speed. To further enhance the multi-case PINN, we explored two strategies. First, incorporating coordinate parameters relevant to tube geometry, such as distance to wall and centerline distance, as inputs to PINN, significantly enhanced accuracy and reduced computational burden. Second, the addition of extra loss terms, enforcing zero derivatives of existing physics constraints in the PINN (similar to gPINN), improved the performance of the Mixed Network and Hypernetwork, but not that of the Modes network. In conclusion, our work identified strategies crucial for future scaling up to 3D, wider geometry ranges, and additional flow conditions, ultimately aiming towards clinical utility. |
| format | Article |
| id | doaj-art-d6a11fb1a3d9485cbc244b448193fcb7 |
| institution | Kabale University |
| issn | 2045-2322 |
| language | English |
| publishDate | 2024-05-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Scientific Reports |
| spelling | doaj-art-d6a11fb1a3d9485cbc244b448193fcb72024-12-29T12:17:25ZengNature PortfolioScientific Reports2045-23222024-05-0114111510.1038/s41598-024-62117-9Strategies for multi-case physics-informed neural networks for tube flows: a study using 2D flow scenariosHong Shen Wong0Wei Xuan Chan1Bing Huan Li2Choon Hwai Yap3Department of Bioengineering, Imperial College LondonDepartment of Bioengineering, Imperial College LondonDepartment of Chemical Engineering, Imperial College LondonDepartment of Bioengineering, Imperial College LondonAbstract Fluid dynamics computations for tube-like geometries are crucial in biomedical evaluations of vascular and airways fluid dynamics. Physics-Informed Neural Networks (PINNs) have emerged as a promising alternative to traditional computational fluid dynamics (CFD) methods. However, vanilla PINNs often demand longer training times than conventional CFD methods for each specific flow scenario, limiting their widespread use. To address this, multi-case PINN approach has been proposed, where varied geometry cases are parameterized and pre-trained on the PINN. This allows for quick generation of flow results in unseen geometries. In this study, we compare three network architectures to optimize the multi-case PINN through experiments on a series of idealized 2D stenotic tube flows. The evaluated architectures include the ‘Mixed Network’, treating case parameters as additional dimensions in the vanilla PINN architecture; the “Hypernetwork”, incorporating case parameters into a side network that computes weights in the main PINN network; and the “Modes” network, where case parameters input into a side network contribute to the final output via an inner product, similar to DeepONet. Results confirm the viability of the multi-case parametric PINN approach, with the Modes network exhibiting superior performance in terms of accuracy, convergence efficiency, and computational speed. To further enhance the multi-case PINN, we explored two strategies. First, incorporating coordinate parameters relevant to tube geometry, such as distance to wall and centerline distance, as inputs to PINN, significantly enhanced accuracy and reduced computational burden. Second, the addition of extra loss terms, enforcing zero derivatives of existing physics constraints in the PINN (similar to gPINN), improved the performance of the Mixed Network and Hypernetwork, but not that of the Modes network. In conclusion, our work identified strategies crucial for future scaling up to 3D, wider geometry ranges, and additional flow conditions, ultimately aiming towards clinical utility.https://doi.org/10.1038/s41598-024-62117-9Physics informed neural networkFluid flowDeep learningHypernetworks |
| spellingShingle | Hong Shen Wong Wei Xuan Chan Bing Huan Li Choon Hwai Yap Strategies for multi-case physics-informed neural networks for tube flows: a study using 2D flow scenarios Scientific Reports Physics informed neural network Fluid flow Deep learning Hypernetworks |
| title | Strategies for multi-case physics-informed neural networks for tube flows: a study using 2D flow scenarios |
| title_full | Strategies for multi-case physics-informed neural networks for tube flows: a study using 2D flow scenarios |
| title_fullStr | Strategies for multi-case physics-informed neural networks for tube flows: a study using 2D flow scenarios |
| title_full_unstemmed | Strategies for multi-case physics-informed neural networks for tube flows: a study using 2D flow scenarios |
| title_short | Strategies for multi-case physics-informed neural networks for tube flows: a study using 2D flow scenarios |
| title_sort | strategies for multi case physics informed neural networks for tube flows a study using 2d flow scenarios |
| topic | Physics informed neural network Fluid flow Deep learning Hypernetworks |
| url | https://doi.org/10.1038/s41598-024-62117-9 |
| work_keys_str_mv | AT hongshenwong strategiesformulticasephysicsinformedneuralnetworksfortubeflowsastudyusing2dflowscenarios AT weixuanchan strategiesformulticasephysicsinformedneuralnetworksfortubeflowsastudyusing2dflowscenarios AT binghuanli strategiesformulticasephysicsinformedneuralnetworksfortubeflowsastudyusing2dflowscenarios AT choonhwaiyap strategiesformulticasephysicsinformedneuralnetworksfortubeflowsastudyusing2dflowscenarios |