RLGC

Repo of the Reinforcement Learning for Grid Control (RLGC) Project.

In this project, we explore to use deep reinforcement learning methods for control and decision-making problems in power systems. We leverage the InterPSS simulation platform (http://www.interpss.org/) as the power system simulator. We develop an OpenAI gym (https://gym.openai.com/) compatible power grid dynamic simulation environment for developing, testing and benchmarking reinforcement learning algorithms for grid control.

NOTE: RLGC is under active development and may change at any time. Feel free to provide feedback and comments.

Environment setup

To run the training, you need python 3.5 or above and Java 8. Unix-based OS is recommended. We suggest using Anaconda to create virtual environment from the yaml file we provided.

• To clone our project

  git clone https://github.com/RLGC-Project/RLGC.git
  

• To create the virtual environment
  In case you like to use our development environment, we have provided environment. yml file

  cd RLGC
  conda env create -f environment.yml
  

  or you can create your own environment. The main dependent modules/libs include gym, tensorflow, py4j, numpy, matplotlib, stable-baselines,jupyter-notebooks

  cd RLGC    
  conda env create --name <your-env-name>  
  

  If you get errors about OpenAI gym you probably need to install cmake and zlib1g-dev. For example on Ubuntu machine, do the following command.

  sudo apt-get upgrade
  sudo apt-get install cmake
  sudo apt-get install zlib1g-dev
  

  After creating environment , you can activate the virtual environment and do development under this environment.

• To activate virtual environment

  source activate <your-env-name>  
  

• To deactivate virtual environment

  source deactivate
  

Training

• With the RLGCJavaSever version 0.80 or newer and grid environment definition version 5 (PowerDynSimEnvDef_v5.py) or newer, users don't need to start the java server explicitly. The server will be started automatically when the grid environment PowerDynSimEnv is created.
• To launch the training, you need first activate the virtual environment. Then run the training scripts under the folder

source activate <your-env-name> 
cd RLGC/examples/IEEE39_load_shedding/  
python trainIEEE39LoadSheddingAgent_discrete_action.py 

During the training the screen will dump out the training log. After training, you can deactivate the virtual environment by

source deactivate

Check training results and test trained model

Two Jupyter notebooks (with Linux and Windows versions-- directory paths are specified differently) are provided as examples for checking training results and testing trained RL model.

Customize the grid environment for training and testing

If you want to develop a new grid environment for RL training or customize the existing grid environment (e.g. IEEE 39-bus system for load shedding), the simplest way is through providing your own cases and configuration files.

When you open trainIEEE39LoadSheddingAgent_discrete_action.py you will notice the following codes:

case_files_array =[]
case_files_array.append(repo_path + '/testData/IEEE39/IEEE39bus_multiloads_xfmr4_smallX_v30.raw')
case_files_array.append(repo_path + '/testData/IEEE39/IEEE39bus_3AC.dyr')

....
# configuration files for dynamic simulation and RL
dyn_config_file = repo_path + '/testData/IEEE39/json/IEEE39_dyn_config.json'
rl_config_file = repo_path + '/testData/IEEE39/json/IEEE39_RL_loadShedding_3motor_2levels.json'

env = PowerDynSimEnv(case_files_array,dyn_config_file,rl_config_file, jar_path, java_port)

They are to specify the cases and configuration files for dynamic simulation and RL training. You can develop your environment by following these examples. Since PowerDynSimEnv is defined based on OpenAI Gym environment definition, once the environment is created, you can use it like other Gym environments, and seamlessly interface it with RL algorithms provided in OpenAI baselines or Stable baselines

Citation

If you use this code please cite it as:

@article{huang2019adaptive,
  title={Adaptive Power System Emergency Control using Deep Reinforcement Learning},
  author={Huang, Qiuhua and Huang, Renke and Hao, Weituo and Tan, Jie and Fan, Rui and Huang, Zhenyu},
  journal={IEEE Transactions on Smart Grid},
  year={2019},
  publisher={IEEE}
}

Communication

If you spot a bug or have a problem running the code, please open an issue.

Please direct other correspondence to Qiuhua Huang: qiuhua DOT huang AT pnnl DOT gov