TY  - EJOU
AU  - Asadi, Saeed 
AU  - Naeini, Hajar Kazemi 
AU  - Hassanlou, Delaram 
AU  - Pishahang, Abolhassan 
AU  - Najafabadi, Saeid Aghasoleymani 
AU  - Sharifi, Abbas 
AU  - Ahmadi, Mohsen 

TI  - AI-Powered Digital Twin Frameworks for Smart Grid Optimization and Real-Time Energy Management in Smart Buildings: A Survey
T2  - Computer Modeling in Engineering \& Sciences

PY  - 2025
VL  - 145
IS  - 2
SN  - 1526-1506

AB  - The growing energy demand of buildings, driven by rapid urbanization, poses significant challenges for sustainable urban development. As buildings account for over 40% of global energy consumption, innovative solutions are needed to improve efficiency, resilience, and environmental performance. This paper reviews the integration of Digital Twin (DT) technologies and Machine Learning (ML) for optimizing energy management in smart buildings connected to smart grids. A key enabler of this integration is the Internet of Things (IoT), which provides the sensor networks and real-time data streams that fee/d DT–ML frameworks, enabling accurate monitoring, forecasting, and adaptive control. Through this synergy, DT–ML systems enhance energy prediction, occupant comfort, and automated fault detection, while also supporting broader sustainability goals. The review examines recent advances in DT–ML energy systems, with attention to enabling technologies such as IoT sensor networks, building energy management systems, edge–cloud computing, and advanced analytics. Key challenges including data interoperability, cybersecurity, scalability, and the need for standardized frameworks are critically discussed, along with emerging solutions such as federated learning and blockchain. Special focus is given to human-centric digital twin frameworks that integrate user comfort and behavioral adaptation into energy optimization strategies. The findings suggest that DT–ML integration, enabled by IoT sensor networks, has the potential to significantly reduce energy consumption, lower operational costs, and improve resilience in urban infrastructures. The paper concludes by outlining future research priorities, including decentralized learning models, universal data standards, enhanced privacy protocols, and expanding digital twin applications for distributed renewable energy resources.
KW  - Digital twin; machine learning; smart grid; smart buildings; energy optimization; IoT; real-time monitoring; sustainability

DO  - 10.32604/cmes.2025.070528