AIPlan4Grid

This repository contains the code of the AIPlan4Grid agent, which is an agent that uses the UnifiedPlanning optimization framework in order to solve powergrid control problem thanks to the Grid2Op modelisation.

Installation of the package

Python 3.10.0 is required.

You can install the package with:

python -m pip install git+https://github.com/aiplan4eu/AIPlan4Grid.git@<version>

Where <version> is the version of the released package you want to install.

Or you can also install the package from the wheel file located in the dist folder after building the package. See the Building the package section for more information.

Main usage

CLI

python -m plan4grid -h
usage: python -m plan4grid [-h] -e <env_name> -s <scenario_id> [-c <config_file>] [-d] [--save]

Run the AIPlan4Grid agent on the given environment with the given scenario. If no configuration file is given, the default configuration file will be used.

optional arguments:
  -h, --help            show this help message and exit
  -e ENV_NAME, --env-name ENV_NAME
                        Name of the environment to use.
  -s SCENARIO_ID, --scenario-id SCENARIO_ID
                        ID of the scenario to use.
  -c CONFIG_FILE, --config-file CONFIG_FILE
                        Configuration file path. If not given, the default configuration file will be used.
  -d, --debug           Debug mode.
  --save                Save the results.

Configuration file

A configuration file can be provided to the agent. If no configuration file is provided, the default configuration file will be used. The default configuration file is parameters.ini.

It looks like this:

[Parameters]
tactical_horizon = 1 # The number of steps to look ahead in the future
strategic_horizon = 288 # The number of time steps over which the agent is operated
solver = enhsp # The solver to use (currently only enhsp is supported)
noise = False # Whether to add noise to the observations or not
test = False # Whether to run the agent in test mode or not

Notebook

You can also run the agent from a notebook using the Launcher class. See the notebooks folder and the API documentation for more information.

Development requirements

In order to develop the package, you need to install the required dependencies for development. Moreover, it is highly recommended to use a virtual environment.

pip install -r requirements/requirements.txt

Building documentation

Documentation can be built using sphinx and by installing required dependencies.

python -m pip install -r requirements/requirements_docs.txt
python -m sphinx docs/source docs/build/

Building the package

Build, test and install the package with the following command:

python -m pip install -r requirements/requirements_build.txt
cd scripts
chmod +x build_pkg.sh
./build_pkg.sh

Note: if you want to update the version number of the package, you'll have to do it in the setup.py file AND in the __init__.py file of the plan4grid folder (the __version__ variable) before building the package.

Grid2Viz

After running the agent, you can visualize the actions he took and the grid states using the Grid2Viz tool.

cd scripts
chmod +x visualize.sh
./visualize.sh <results_path> <g2op_env_path>

Where <results_path> is the path to the results directory (stored in a folder called "results") and <g2op_env_path> is the path to the environment data directory stored by Grid2Op in a folder called "data_grid2op".