AI-Based Solar Energy Optimization

Tools & Technologies Used in this Project

✅ Best Performing Model: 15-minute interval forecasting
⚠️ Worst Performing Model: 24-hour interval forecasting
Overall Forecasting Performance
🎯 Main Goal of the Project

To design and evaluate AI-based solar energy forecasting models, integrating them into smart grid systems for improved energy management. The project aims to compare forecasting performance, explore optimization techniques, and contribute to sustainable energy deployment using intelligent data-driven methods.

🖥️ Frontend

  • HTML5: Page structure and layout
  • CSS3: Styling and responsiveness
  • Bootstrap 5: UI components and grid
  • JavaScript: DOM interaction and logic
  • Chart.js: Interactive visualizations

⚙️ Backend & AI

  • Python 3.x: Backend scripting and model control
  • Flask (optional): Serve predictions via API
  • TensorFlow/Keras: Deep learning with LSTM
  • scikit-learn: Preprocessing and validation
  • NumPy & Pandas: Data manipulation
  • pvlib: Solar irradiance estimation (GHIcs)

📊 Data Handling

  • Dataset: 10-second interval GHI measurements
  • Cleaning: NaN handling, outlier removal
  • Features: kcs, seasonal sinusoidal features, derivatives
  • Validation: MAE, RMSE, nRMSE, R² metrics

🌐 Hosting & Integration

  • Flask API: Serve model results to UI
  • GitHub: Source control and versioning
  • MySQL: Real-time database (optional)
Project Presented By

Siva Haritha, Intern - NIC Division

Under the mentorship of Shri. Pankaj Bajaj, NIC Senior Director (Scientist-F)