Parameteric DeepONet for modeling structural dynamics

This repository is the code and data for the manuscript: Parameter estimation of structural dynamics with neural operators enabled surrogate modeling.

Installation

Deep operator networks (DeepONet) is based on the DeepXDE. Install DeepXDE by:

pip install deepxde

The implementations are based on Ubuntu 20.04, Python 3.8, and PyTorch-Cuda 11.6.

Data

Please refer to ./data/ folder.

Train the foward surrogate model

Set the training configurations via .yaml config file, and start training by the run.sh.

sh run.sh

Evaluate

Experiments of Case1 and Case2 are in the ./experiments/ folder.

Case 1

SDOF Response prediction, see example in case1b_forward.ipynb.

Parameter estimation, see example inin case1b_inverse.ipynb.

Case 2

MDOF response prediction, see example in case2_forward_params_a.ipynb.

Damage length estimation, see example in case2_inverse_params_a.ipynb.

Damage shape estimation, see example in case2_inverse_params_b.ipynb.

(Set the suitable parameterization code in inverse_net.py and script.py)

Acknowledgement

Our code is partially based on:
https://github.com/lululxvi/deepxde
https://github.com/adler-j/learned_gradient_tomography
https://github.com/csiro-mlai/fno_inversion_ml4ps2021