Strategies for multi-case physics-informed neural networks for tube flows: a study using 2D flow scenarios
2024

Improving Neural Networks for Fluid Flow in Tubes

Sample size: 61 publication 10 minutes Evidence: high

Author Information

Author(s): Wong Hong Shen, Chan Wei Xuan, Li Bing Huan, Yap Choon Hwai

Primary Institution: Imperial College London

Hypothesis

Can multi-case physics-informed neural networks (PINNs) optimize fluid flow simulations in tube-like geometries?

Conclusion

The study found that the Modes network architecture outperformed others in accuracy and efficiency for simulating fluid flows in tube-like structures.

Supporting Evidence

  • The Modes network showed superior performance in terms of accuracy and computational speed.
  • Incorporating tube-specific coordinate parameters significantly enhanced accuracy.
  • The multi-case PINN approach allows for quick generation of flow results in unseen geometries.

Takeaway

This study shows how special computer programs can help us understand how fluids move in tubes, which is important for things like blood flow in our bodies.

Methodology

The study compared three neural network architectures for fluid flow simulations in 2D tube-like geometries, using a series of idealized stenotic tube flows.

Limitations

The algorithms are not yet ready for clinical applications and are limited to 2D flows within a specific geometric parameter space.

Statistical Information

P-Value

p<0.05

Statistical Significance

p<0.05

Digital Object Identifier (DOI)

10.1038/s41598-024-62117-9

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication