🔋 Industrial Power Grid Environment
An autonomous Reinforcement Learning environment designed to simulate an industrial power grid. The goal is to balance energy supply from multiple sources (Solar, Wind, Battery) to meet factory demand while minimizing the total carbon footprint.
🚀 Quick Start
The simplest way to interact with the environment is using the Gymnasium wrapper provided in the repository:
from gymnasium_wrapper import PowerGridGymEnv
# Initialize the environment (Ensure server is running on port 8001)
env = PowerGridGymEnv(endpoint="http://localhost:8001")
obs, info = env.reset()
print(f"Initial State: {obs}")
# Take a step with a random action (0: Solar, 1: Wind, 2: Battery)
action = env.action_space.sample()
obs, reward, terminated, truncated, info = env.step(action)
print(f"Action Taken: {action} | Reward: {reward}")
🛠️ Installation & Setup
1. Install Dependencies
Ensure you have your virtual environment active, then run:
Bash
pip install -r requirements.txt
2. Run the Environment ServerStart the FastAPI/OpenEnv server to host the simulation:Bash# Windows (PowerShell)
$env:PYTHONPATH = "."
python -m server.app --port 8001
# Linux/Mac
PYTHONPATH=. python3 -m server.app --port 8001
3. Run the AI Stability Test
In a second terminal, run the training script to verify the RL loop:
Bash
python train.py
📊 Environment Details
Action Space: Discrete(3)
Value Action Description
0 Toggle Solar Activates/Deactivates Solar panel arrays
1 Toggle Wind Activates/Deactivates Wind turbine clusters
2 Toggle Battery Switches between Charging/Discharging states
Observation
Space: Box(3)
Index Feature Unit
0 Demand Met Boolean (1.0 = Yes, 0.0 = No)
1 Carbon FootprintMetric Tons (Lower is better)
2 Grid Stability 1.0 = Stable, 0.0 = Unstable
Reward Function
The agent is rewarded for maintaining a stable grid while reducing emissions:
Success: $+1.0$ (Demand met with 0 carbon)Sustainability Penalty: Reward is scaled by carbon footprint: $1.0 - (Carbon \times 0.5)$Failure: $-1.0$ (Grid blackout/demand not met)
🐳 Docker Support
Building the Image
Bash
docker build -t industrial_power_grid-env:latest -f server/Dockerfile .
Deploying to Hugging Face Spaces
Bash
openenv push --repo-id your-username/industrial-power-grid --private
📡 API Documentation
Once the server is running, you can explore the interactive API docs at:
http://localhost:8001/docs
## Project Structure
industrial_power_grid/ ├── .dockerignore # Docker build exclusions ├── init.py # Module exports ├── README.md # This file ├── openenv.yaml # OpenEnv manifest ├── pyproject.toml # Project metadata and dependencies ├── uv.lock # Locked dependencies (generated) ├── client.py # IndustrialPowerGridEnv client ├── models.py # Action and Observation models └── server/ ├── init.py # Server module exports ├── industrial_power_grid_environment.py # Core environment logic ├── app.py # FastAPI application (HTTP + WebSocket endpoints) └── Dockerfile # Container image definition