Multi-Objective Reinforcement Learning for Power Grid Topology Control

This repository contains a collection of modules for implementing Multi-Objective Reinforcement Learning (MORL) algorithm specifically designed for multi-objective power grid topology control. This framework bases on morl_baselines for the MORL part and on Grid2Op for the power system environment.

Table of Contents

• Installation
• Modules
  • ols_DOL_exe.py
  • MO_PPO.py
  • GridRewards.py
  • CustomGymEnv.py
  • EnvSetup.py
  • MO_PPO_train_utils.py
  • MORL_analysis_utils.py
  • Grid2op_eval.py
  • env_start_up.py
• Usage
• Contributing
• License

Installation

Modules

The modules are divided into source and scripts. Source modules are sperated into environment, agent, wrapper and utils.

ols_DOL_exe.py

Starts and proceeds the experiments including DOL and MOPPO Training and Evaluation.

MO_PPO.py

Contains the implementation of the Multi-Objective Proximal Policy Optimization (MO-PPO) algorithm, based on the MORL_baseline package

GridRewards.py

Contains the implementation for calculating grid-based rewards.

Classes:

• GridRewards: Calculates rewards based on a grid of metrics.

CustomGymEnv.py

Defines a custom Gym environment for MORL experiments.

EnvSetup.py

Utility for setting up the custom Gym environment.

MO_PPO_train_utils.py

Contains utility functions for training MO-PPO.

MORL_analysis_utils.py

Contains utility functions for analyzing MORL experiments.

case_studies.py

Contains the analysis script and plotting for the case studies

Grid2op_eval.py

Contains the evaluation script for Grid2Op environment.

env_start_up.py

Sets up the environment for power grid topology control experiments.

Contributing

Contributions are welcome Please create an issue or submit a pull request for any changes.

License

This work is licensed under a