Renewables_Scenario_Gen_GAN-TensorFlow

Unofficial implementation of Renewables Scenario Generation GAN (Chen et al., IEEE 2018) in PyTorch 2.

Y. Chen, Y. Wang, D. Kirschen and B. Zhang, "Model-Free Renewable Scenario Generation Using Generative Adversarial Networks," in IEEE Transactions on Power Systems, vol. 33, no. 3, pp. 3265-3275, May 2018, doi: 10.1109/TPWRS.2018.2794541

Original Codebase: https://github.com/chennnnnyize-zz/Renewables_Scenario_Gen_GAN

Data Set Reference

• Solar Data: https://www.nrel.gov/grid/solar-power-data.html
• Wind Data: https://www.nrel.gov/grid/wind-integration-data.html

Version Notes

The model was implemented and tested using Python==3.11.9. Further, the following modules were utilized (see Requirements File):

matplotlib==3.10.0
numpy==2.2.1
scipy==1.15.0
torch==2.5.1

Usage

To conduct the experiments, the easiest way to get started is by cloning this repository and use one of the notebooks (solar/wind).

Alternatively, run it from the terminal.

py train.py --data=datasets/solar.csv --label=datasets/solar_label.csv --epochs=5000 --batch_size=32 --learning_rate=1e-4

Results

The recreated experiments demonstrate the GAN's capability to resemble the data distribution and leverage the provided labels for conditioned generation of certain scenarios, even after short training trials. Obviously, the training can be extended and parameters adapted to improve the GAN's fidelity.

Solar Generation

The plots present synthesized solar power generation scenarios arranged by ordinal label values.

Wind Generation

The plots present synthesized wind power generation scenarios arranged by ordinal label values.