MPC-driven building energy management for privacy and zero-carbon trade-off optimization using energy storage as physical noise

As the global demand for renewable energy surges, the need for detailed electricity usage data is critical for effective power management. Building Energy Management Systems (BEMS) are thus confronted with dual objectives: optimizing energy consumption and safeguarding user privacy. This paper introduces a Model Predictive Control (MPC)-based BEMS designed to achieve a balance between zero-carbon emissions and privacy protection. The proposed system employs an MPC algorithm to manage real-time energy flows through residential energy storage devices, such as batteries. This approach not only diminishes households' reliance on the power grid and maximizes carbon emission reduction but also leverages the physical noise by storage devices to mask actual electricity consumption curves and patterns, thereby enhancing privacy protection. Comprehensive simulation experiments indicate that the MPC algorithm substantially surpasses conventional scheduling methods in terms of achieving a balance between carbon emission reduction and enhancing privacy protection. Notably, it achieves a 42.6 %–50.9 % reduction in energy consumption and a 19.7 % reduction in carbon emissions compared to sophisticated algorithms like Deep Q-Network (DQN). Furthermore, the MPC algorithm enhances privacy by ensuring a cosine similarity of up to 96.5 % between the optimized and original electricity usage patterns, demonstrating robust privacy safeguards. Sensitivity analysis further validates the algorithm's robustness and adaptability across diverse environmental scenarios. The study concludes that the MPC-based BEMS offers both theoretical and practical advantages for efficient buidling energy management and user privacy protection, positioning it as a promising solution for future zero-carbon and privacy-conscious energy systems.