ThermoPINNs

This repository includes the Python package ThermoPINNs, which uses Physics-Informed Neural Networks (PINNs) to solve thermal problems in single-track laser powder bed fusion (LPBF).

Source Code

To access the source code for ThermoPINNs, see the folder PINNs_code/ThermoPINNs/. You can modify the config.yaml file and run test.py to train a PINNs model. A manual for each parameter in config.yaml is provided in PINNs_code/ThermoPINNs/config_function.py.

Dependencies

For package dependencies, refer to Python on Euler. In addition to this environment, install sobol-seq and pyDOE using:

pip install pyDOE sobol-seq

Modifications for New Problems

For new problems, such as different materials or process conditions, modify PINNs_code/ThermoPINNs/Paper_Equation_param.py as needed.