Cost-oriented-Generative-Model

This repository contains the code and related materials for the paper “Cost-oriented Scenario Generation for Power Systems Under Uncertainty”, which has been submitted to IEEE Transactions on Smart Grid.

Authors: Yangze Zhou, Yihong Zhou, Daniel Kirschen, Thomas Morstyn, and Yi Wang.

Corresponding author: Yi Wang (yiwang@eee.hku.hk)

Data

Data can be found in Google Drive Folder

Environment

We recommend using Conda to manage the Python environment for this project.

Create the environment from the provided file:

conda env create -f environment.yml
conda activate Meta_DFL

How to Run

Training Setting	Model	How to Run
Standard generative model	VAE	Run the notebooks forecasting_VAE_joint.ipynb and forecasting_VAE_separate.ipynb
	GAN	Run the notebooks forecasting_GAN_joint.ipynb and forecasting_GAN_separate.ipynb
	Diffusion	Run the notebooks forecasting_diffusion_joint.ipynb and forecasting_diffusion_separate.ipynb
Cost-oriented generative model	VAE	Run the commands python main_VAE_joint.py and python main_VAE_separate.py
	GAN	Run the commands python main_GAN_joint.py and python main_GAN_separate.py
	Diffusion	Run the commands python main_diffusion_joint.py and python main_diffusion_separate.py

Results

Because the forecasting model parameters are too large to be uploaded to GitHub, they are provided separately via Google Drive.

Please note that this repository currently contains only partial experimental results. For the complete version, including the full model parameters and related files, please visit:

Google Drive Folder