README.md

Reinforcement Learning for Water Distribution System Control

Setup

1. Clone and install the environment

git clone https://github.com/Daveonwave/gym4ReaL/
cd gym4ReaL/
pip install -e . --break-system-packages
cd ..

2. Install dependencies

pip install -r requirements.txt --break-system-packages

3. Move reward wrapper to environment

cp algorithms/reward_scaling_wrapper.py gym4ReaL/gym4real/envs/wds/

Training

Run algorithms from the project root:

# Example train 
python3 algorithms/PPO.py
python3 algorithms/TRPO.py
python3 algorithms/DQN.py
python3 algorithms/A2C.py

Monitoring Training

tensorboard --logdir logs/

Environment Details

Observation space (27 dimensions):

• Tank water levels (2)
• Node pressures (22)
• Demand forecast values (2)
• Time encoding (1)

Action space (Discrete, 4 actions):

• Pump on/off combinations for the 2 pumps

Reward: Based on maintaining pressure bounds and minimizing tank overflow

Cloned environment

gym4ReaL - Gymnasium environment for water networks