⚡ Smart Grid AI System

A production-grade, end-to-end Smart Grid AI application designed for real-time monitoring, load forecasting, and fault detection. The system leverages the U.S. EIA API for live electricity demand data and employs advanced machine learning models to ensure grid stability and efficiency.

---

🚀 Key Features

• Live Monitoring: Real-time dashboard with status indicators (Normal, Overload, Under-utilized).
• ML Forecasting: Predictive load modeling using Random Forest Regressors with cyclical feature engineering.
• Anomaly Detection: Automated fault detection via Isolation Forest, identifying grid irregularities in real-time.
• Intelligent Optimization: Rule-based demand-response suggestions for load shifting and peak shaving.
• Voice Integration: Built-in voice assistant for hands-free status reports.
• Data Portability: Instant CSV export for historical analysis.
• Modern UI: High-performance, dark-mode dashboard powered by Chart.js.

---

🏗️ Architecture

The system is built on a modular architecture:

• Data Layer: Handles communication with the U.S. Energy Information Administration (EIA) API, with a robust fallback to synthetic data generation.
• Preprocessing Engine: Implements complex feature engineering, including sinusoidal transformations for temporal data and rolling-window statistics.
• Model Service: Manages training and inference for an ensemble of models (Random Forest + Isolation Forest).
• Web API: A Flask-based RESTful interface serving processed data and model predictions.
• Frontend: A responsive single-page application (SPA) providing visual insights.

---

📂 Project Structure

Smart-Grid-AI-System/
├── app.py                  # Core Flask server & API Layer
├── train_model.py          # ML Training Pipeline
├── model.pkl               # Serialized Model Bundle (generated)
├── requirements.txt        # System Dependencies
├── README.md               # Documentation
├── static/
│   ├── script.js           # Dashboard Logic & Charting
│   └── style.css           # Premium Dark-Mode Styling
├── templates/
│   └── index.html          # Main UI Structure
└── utils/
    ├── data_fetch.py       # API Integration & Mock Logic
    └── preprocessing.py    # Feature Engineering & Normalization

---

🛠️ Installation & Setup

1. Environment Setup

# Clone the repository
git clone <repository-url>
cd Smart-Grid-AI-System

# Create and activate virtual environment
python -m venv .venv
source .venv/bin/activate  # Windows: .venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

2. API Configuration (Optional)

The system works out-of-the-box with mock data. To use live data, obtain an EIA API Key and set it:

export EIA_API_KEY="your_api_key_here"  # Windows: set EIA_API_KEY=your_api_key_here

3. Initialize Models

python train_model.py

4. Launch Dashboard

python app.py

Visit http://localhost:5000 in your browser.

---

📊 Machine Learning Insights

Load Forecaster

• Algorithm: Random Forest Regressor
• Features: Hour (sin/cos), Day of Week, Month, Weekend flag, Peak Hour flags.
• Performance: Consistently outperforms linear baselines by capturing non-linear daily patterns.

Anomaly Detector

• Algorithm: Isolation Forest
• Logic: Analyzes residuals between actual demand and predicted load. High residuals are flagged as potential grid faults or sensor errors.
• Contamination: Dynamically adjusted based on historical variance.

---

🔮 Future Roadmap

• Multi-Region Support: Real-time switching between different ISO/RTO regions.
• Weather Integration: Correlate load forecasts with real-time temperature and humidity data.
• Deep Learning Migration: Implementation of LSTM/GRU networks for improved time-series forecasting.
• Containerization: Full Docker support for seamless cloud deployment.

---

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

---

Developed with ❤️ for Smart Energy Management.