/gym-anm

Design Reinforcement Learning environments that model Active Network Management (ANM) tasks in electricity distribution networks.

Primary LanguagePythonMIT LicenseMIT

Gym-ANM

Documentation Status codecov CI (pip) CI (conda) License: MIT

gym-anm is a framework for designing reinforcement learning (RL) environments that model Active Network Management (ANM) tasks in electricity distribution networks. It is built on top of the OpenAI Gym toolkit.

The gym-anm framework was designed with one goal in mind: bridge the gap between research in RL and in the management of power systems. We attempt to do this by providing RL researchers with an easy-to-work-with library of environments that model decision-making tasks in power grids.

Paper: Gym-ANM: Reinforcement Learning Environments for Active Network Management Tasks in Electricity Distribution Systems

Key features

  • Very little background in electricity systems modelling it required. This makes gym-anm an ideal starting point for RL students and researchers looking to enter the field.
  • The environments (tasks) generated by gym-anm follow the OpenAI Gym framework, with which a large part of the RL community is already familiar.
  • The flexibility of gym-anm, with its different customizable components, makes it a suitable framework to model a wide range of ANM tasks, from simple ones that can be used for educational purposes, to complex ones designed to conduct advanced research.

Documentation

Documentation is provided online at https://gym-anm.readthedocs.io/en/latest/.

Installation

Requirements

gym-anm requires Python 3.7+ and can run on Linux, MaxOS, and Windows.

We recommend installing gym-anm in a Python environment (e.g., virtualenv or conda).

Using pip

Using pip (preferably after activating your virtual environment):

pip install gym-anm

Building from source

Alternatively, you can build gym-anm directly from source:

git clone https://github.com/robinhenry/gym-anm.git
cd gym-anm
pip install -e .

Example

The following code snippet illustrates how gym-anm environments can be used. In this example, actions are randomly sampled from the action space of the environment ANM6Easy-v0. For more information about the agent-environment interface, see the official OpenAI Gym documentation.

import gym
import time

env = gym.make('gym_anm:ANM6Easy-v0')
o = env.reset()

for i in range(100):
    a = env.action_space.sample()
    o, r, done, info = env.step(a)
    env.render()
    time.sleep(0.5)  # otherwise the rendering is too fast for the human eye.

The above code would render the environment in your default web browser as shown in the image below: alt text

Additional example scripts can be found in examples/.

Testing the installation

All unit tests in gym-anm can be ran from the project root directory with:

python -m tests

Citing the project

To cite this project in publications, cite the original paper:

@misc{henry2021gymanm,
      title={Gym-ANM: Reinforcement Learning Environments for Active Network Management Tasks in Electricity Distribution Systems}, 
      author={Robin Henry and Damien Ernst},
      year={2021},
      eprint={2103.07932},
      archivePrefix={arXiv},
      primaryClass={cs.LG}
}

Maintainers

gym-anm is currently maintained by Robin Henry.

License

This project is licensed under the MIT License - see the LICENSE.md file for details.