A modular cyber-physical architecture for residential microgrids with slot-based, hardware-constrained energy management

Advanced energy management systems (EMSs) show clear potential for complex multi-energy systems, but a practical gap still limits their deployment. Much of the current literature relies on idealized simulations, assumes perfect forecasting, and uses centralized architectures that do not fully account for the rigid operating constraints of commercial hardware. To address this gap, this paper presents a modular Cyber–Physical System (CPS) framework for real-world, hardware-constrained energy management. Based on a decentralized Message Queuing Telemetry Transport (MQTT) architecture, the proposed CPS coordinates physical assets and predictive cyber modules across different software platforms. Within this framework, a slot-based EMS is developed to respect the control limits of a commercial hybrid inverter by replacing standard continuous optimization with adaptive time-slot classification. To evaluate the scalability of the CPS, a virtual reformed-methanol fuel cell and its equivalent pricing model are integrated into the live control loop without system-wide reconfiguration. The framework is deployed on an active residential microgrid in Denmark, where the mirroring module is validated against the physical hardware with correlations of 0.94–0.99. All automated strategies are then evaluated on this hardware-validated virtual testbed, driven by the measured field data. Compared with the measured manual operation, the proposed EMS reduces operating costs by 10.12% and extends the estimated blackout self-support time from 12.7 h to 28.8 h. It also remains robust as forecast error grows, keeping its advantage even at ±60% error, while the finer continuous schedule degrades faster. These results show that the proposed low-resolution, hardware-constrained EMS can translate theoretical optimization into practical, hardware-executable operation.