Digital twin assisted real-time energy management system for smart homes

The increasing integration of renewable energy sources into modern power systems has amplified the need for efficient, real-time energy management solutions at the household level. Smart homes, equipped with distributed generation units, energy storage systems, electric vehicles and flexible loads, are emerging as active participants in decentralized energy networks. However, the intermittent nature of renewable generation, dynamic energy pricing, and real-time uncertainties in user behavior pose significant challenges for conventional home energy management systems. This paper presents a digital twin-assisted, edge-based real-time home energy management system designed to optimize energy consumption and operational costs in smart homes. The proposed system integrates a real-time digital twin of the home’s energy ecosystem with machine learning-based forecasting models and edge computing infrastructure. Long short-term memory networks are employed to predict key parameters such as photovoltaic generation, EV availability, and appliance demand, while the digital twin continuously updates forecasts and simulations based on real-time data. The proposed framework was evaluated through comprehensive simulations using 15-min and 1-min scheduling resolutions. Results demonstrate significant improvements in forecasting accuracy, operational cost reduction, and system responsiveness compared to conventional methods. Specifically, the proposed framework achieved up to 10% cost savings under a 15-min resolution and further improvements with a 1-min resolution, highlighting the benefits of fine-grained, real-time control. The synergy between digital twins, edge computing, and machine learning offers a scalable, privacy-preserving, and responsive energy management framework for future smart homes, contributing to enhanced grid resilience and sustainable energy systems.