A synergistic approach using digital twins and statistical machine learning for intelligent residential energy modelling

The growing need for energy efficiency in buildings has driven significant improvements in digitalisation and intelligent energy management. Traditional energy management techniques often fail to address dynamic energy demands and user preferences, leading to inefficiencies and increased costs. This paper proposes a framework that integrates Digital Twin (DT) systems with Artificial Intelligence (AI) algorithms for intelligent building energy consumption assessment by developing real-time virtual twin representations. Unlike conventional DT models used mainly for visualization or simulation, the proposed approach leverages DTs as adaptive, data-driven decision-making tools that evolve through continuous IoT sensor feedback. This dynamic representation of physical systems enables real-time energy optimization and facilitates intelligent control to enhance both efficiency and sustainability. The system categorizes buildings into energy-efficient and non-energy-efficient groups with an accuracy of 98% by leveraging IoT sensor data, along with Random Forest, Deep Neural Networks, Long Short-Term Memory networks, and Bidirectional Long Short-Term Memory networks. The framework encompasses comprehensive data preprocessing, feature engineering, and the implementation of cutting-edge AI techniques, highlighting the transformative potential of this integrated approach. The results, illustrated through various graphical representations, demonstrate the critical role of DT and AI in optimising energy management, minimising waste, and driving sustainability in industrial and urban environments. Confusion matrices and performance metric graphs reveal that the Random Forest model outperforms other techniques. Meanwhile, training curves and feature importance visualizations provide insights into model behaviour and key factors influencing energy efficiency. This research underscores the significance of combining real-time DT environments with intelligent learning models to address modern energy efficiency challenges and support the development of adaptive, sustainable building systems.