Energy Consumption Disaggregation and Forecasting
Published: GitHub
Overview
Built time-series forecasting application for CS department energy consumption using fine-tuned LSTM, achieving 95% accuracy for 3-day predictions.
Key Features
- Leveraged Graph Signal Processing and GFT for appliance-level energy disaggregation with 90%+ identification accuracy
- Selected for display during college-level Demo-Day by the Entrepreneurship Development Cell of TCE
- Designed to efficiently connect users with electrical energy prediction for both commercial and industrial applications
Technologies
LSTM, Graph Signal Processing, Time-Series Analysis, Deep Learning