Spaces:
Sleeping
Sleeping
metadata
title: Smart Grid Environment Server
emoji: 🎯
colorFrom: yellow
colorTo: blue
sdk: docker
pinned: false
app_port: 8000
base_path: /web
tags:
- openenv
⚡ Smart Grid OpenEnv Environment
A realistic reinforcement learning environment that simulates energy distribution in a smart electrical grid powered by renewable sources.
Built for the Meta x PyTorch OpenEnv Hackathon, this environment enables AI agents to learn how to balance demand, renewable generation, and battery storage.
🌍 Problem Overview
Modern power grids must handle:
- Fluctuating energy demand
- Intermittent solar and wind generation
- Limited battery storage
The goal is to optimally distribute energy across regions while minimizing:
- Unmet demand
- Energy waste
- Battery misuse
🧠 Environment Design
⏱ Time-based Simulation
- Each episode = 24 timesteps (hours)
- Demand and generation vary dynamically over time
📦 Observation Space
Feature Description
------------------ --------------------------
hour Current timestep (0--23)
demand_r1 Demand in Region 1
demand_r2 Demand in Region 2
demand_r3 Demand in Region 3
solar_generation Solar power available
wind_generation Wind power available
battery_level Current battery storage
battery_capacity Maximum battery capacity
🎯 Action Space
The agent must output:
supply_r1, supply_r2, supply_r3, charge_battery
🧩 Tasks
- Easy: balanced_grid_easy
- Medium: solar_management, wind_uncertainty
- Hard: peak_demand, full_grid_challenge
🚀 Running Locally
pip install -r requirements.txt uvicorn server.app:app --reload
🐳 Docker
docker build -t smart_grid_env .
docker run -p 8000:8000 smart_grid_env
🤖 Inference
python inference.py
📁 Project Structure
smart_grid/
├── client.py
├── models.py
├── openenv.yaml
├── inference.py
├── server/
├── tasks/
🏁 Goal
Train an AI agent that balances supply, demand, and storage efficiently.