Smart Grids
Smart Grids
Motivation
Smart grids (SG) rely on efficient use of information and communication technologies to enable automated, reliable, and sustainable electricity production and distribution. A key challenge lies in accurately estimating energy usage, as mismatches between demand and supply increase costs for providers and risk outages. Accurate electricity usage prediction is therefore crucial for designing effective demand response (DR) programs and ensuring sustainability in SG networks.
Goal
The goal of this work is to develop efficient electricity usage prediction methodologies using deep learning techniques, including Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM) networks, and Gated Recurrent Units (GRU). By leveraging these models on real-world electricity consumption data from major U.S. distributors, we aim to enable proactive energy management in smart grids.
Results
After preprocessing publicly available electricity usage data, we trained and validated CNN, LSTM, and GRU models. Experimental results demonstrate that all three models achieve highly accurate predictions, with very low loss (< 0.01), confirming the effectiveness of deep learning for electricity usage forecasting in smart grid systems.
Publications
Listed below are our publications based on this research:
- Anish Roy, R. Sonth, "Efficient Demand-Response Prediction in Smart Grids Using Deep Learning", 12th IEEE Conference on Technologies for Sustainability, (SusTech 2025), Santa Ana, California, April 20-23, 2025. DOI: 10.1109/SusTech63138.2025.11025741