FNO-Stress Strain -- PAPER

Using a neural operator based framework- Fourier Neural Operator FNO to learn the stress strain fields for 2D Composites.

> Stress-Strain Prediction

We use the FNO framework to predcit components of Stress and Strain Tensors $$\hspace{10cm} Stress := \sigma_{xx} ,\sigma_{yy},\sigma_{xy}$$ $$\hspace{10cm} Strain := \varepsilon_{xx} ,\varepsilon_{yy},\varepsilon_{xy}$$

The predicted tensor components are used to measure the equivalent properties like Von-mises stress and Equilvalent Strains.

-Benchmarking

We compare the results of the FNO with ResNet and UNet architectures.

-Dataset

The FE simulations are performed in ABAQUS. Here is the link to the script .The Dataset used are available at FNO-DATASET.It contains: $$E := [Sample, H, W]$$ $$Strains := [Sample,H,W, component]$$ $$Stresses := [Sample, H,W, component]$$

Shield:

This work is licensed under a Creative Commons Attribution 4.0 International License.