Smart Home Energy Management System

A distributed system using Model Context Protocol (MCP) for intelligent home energy management.

Overview

This system uses multiple microservices (MCPs) to analyze, predict, and optimize home energy usage:

Core Services

1. Energy Prediction MCP

   • Predicts energy usage for the next 6 hours
   • Uses machine learning models trained on historical data
   • Endpoint: localhost:8002

2. Energy Bills MCP

   • Analyzes and processes energy billing data
   • Tracks historical energy costs
   • Endpoint: localhost:8001

3. Price Analysis MCP

   • Analyzes energy pricing patterns
   • Predicts optimal usage times
   • Endpoint: localhost:8003

4. Weather MCP

   • Integrates weather data
   • Helps predict weather-based energy needs
   • Endpoint: localhost:8004

Future Tools

5. Device Control MCP

   • Smart device management
   • Automated energy optimization
   • Endpoint: localhost:8005

6. Occupancy Detection MCP

   • Monitors home occupancy patterns
   • Optimizes energy use based on presence
   • Endpoint: localhost:8006

7. Regulations MCP

   • Ensures compliance with energy regulations
   • Monitors energy usage limits
   • Endpoint: localhost:8007

Orchestrator

The MCP Orchestrator (port 8000) coordinates between all services, providing:

• Single service predictions
• Aggregated predictions from multiple services
• Error handling and service health monitoring
• Asynchronous request processing

Installation

1. Clone the repository:

git clone <repository-url>
cd mit-mcp-hackathon

2. Install dependencies for each service:

# For each MCP service
cd <service-name>_mcp
pip install -r requirements.txt

Usage

1. Start the orchestrator:

cd orchestrator_mcp
python server.py

2. Start individual services:

# In separate terminals
cd energy_prediction_mcp
python server.py

cd energy_bills_mcp
python server.py
# ... and so on for other services

API Documentation

Orchestrator Endpoints

1. Single Service Prediction

POST http://localhost:8000/predict/

{
    "service": "energy_prediction",
    "data": {
        "past_24_hours": [1.2, 1.3, ...]  // 24 values
    }
}

2. Aggregated Predictions

POST http://localhost:8000/aggregate-predictions/

{
    "past_24_hours": [1.2, 1.3, ...]  // 24 values
}

Contributing

1. Fork the repository
2. Create your feature branch
3. Commit your changes
4. Push to the branch
5. Create a new Pull Request

License

MIT License

The orchestrator expects MCP services to be running on the following ports:

• Energy Bills MCP: 8001
• Energy Prediction MCP: 8002
• Price Analysis MCP: 8003
• Weather MCP: 8004
• Device Control MCP: 8005
• Occupancy Detection MCP: 8006
• Regulations MCP: 8007

You can modify these in server.py if needed.

Usage

Start the server:

python server.py

The server will run on http://localhost:8000 by default.

API Endpoints

1. Single Service Prediction

POST /predict/

Request body:

{
    "service": "energy_prediction",
    "data": {
        "past_24_hours": [1.2, 1.3, ...]  // 24 values
    }
}

Response:

{
    "predictions": [1.5, 1.6, ...],  // 6-hour predictions
    "service": "energy_prediction"
}

2. Aggregated Predictions

POST /aggregate-predictions/

Request body:

{
    "past_24_hours": [1.2, 1.3, ...]  // 24 values
}

Response:

{
    "aggregated_predictions": [1.5, 1.6, ...]  // Averaged predictions
}

Error Handling

The API returns appropriate HTTP status codes:

• 404: Service not found
• 400: Invalid request data
• 500: Internal server error or service unavailable

Dependencies

• FastAPI
• Uvicorn
• HTTPX
• Pydantic
• Python-dotenv
• Asyncio

Contributing

1. Fork the repository
2. Create your feature branch
3. Commit your changes
4. Push to the branch
5. Create a new Pull Request

License

MIT License