Hierarchical Physics-Embedding Neural Network Framework for 3D Magnetic Modeling of Medium Frequency Transformers
Hierarchical Physics-Embedding Neural Network (HPENN) Framework is a typically designed tool for the 3D Magnetic Modeling of Medium Frequency Transformers (MFTs).
This respository includs all the necessary elements for training, testing, and using HPENN framework.
- Open a terminal and change directory to this program folder:
cd [path-to-folder]- Clone the temperatur conversion repository:
git clone https://github.com/fleaxiao/HPENN.git- Implementation order:
1. data_generation_2D -> train_2D_loss -> test_2D_loss
2. model_combine
3. data_generation_3D -> train_3D_loss -> test_3D_loss
4. Use_HPENN
If you use this code or data in your research, please cite our work as follows:
X. Yang, L. Shu and D. Yang, "Hierarchical Physics-Embedding Neural Network Framework for 3D Magnetic Modeling of Medium Frequency Transformers," in IEEE Transactions on Power Electronics, doi: 10.1109/TPEL.2024.3501573.