Comprehensive design and experimental validation of a hybrid energy management system controller for a residential microgrid using power hardware in-the-loop

Abstract

The large-scale integration of renewable energy sources requires advanced control systems capable of maintaining grid stability and improving energy efficiency under dynamic operating conditions. This study presents the design and experimental validation of a Hybrid Energy Management System for residential microgrids that combines short-term forecasting, Mixed Integer Linear Programming optimization, and real-time control within a unified architecture. The proposed system is implemented and validated through Power Hardware in the Loop testing on a 300 kVA experimental platform at the Smart Grids Test Lab in Ben Guerir, Morocco, using photovoltaic, battery, grid, and load emulators. The forecasting layer generates day-ahead load prediction using multiple algorithms. The Linear Regression model achieves the best accuracy with an R2 of 0.960 and an RMSE of 0.145 kW, while the Random Forest model achieves an R2 of 0.949 and an RMSE of 0.136 kW. Stress and robustness analyses demonstrate that the EMS maintains cost variations below 2 % for forecast deviations of ±10 %. The hybrid control approach achieves an overall cost reduction exceeding 50 percent across multiple scenarios compared with the unoptimized baseline, while maintaining stable real-time performance with an average PLC execution cycle of 120 ms and a maximum response delay below 150 ms. Environmental assessment indicates a 30 to 35 percent decrease in CO2 emissions, equivalent to approximately 12.7–44.7 kg of CO2 saved per day, depending on solar conditions. In addition, the system ensures high user comfort by achieving more than 90 % task completion for flexible loads within preferred time windows while preserving continuous power for critical ones. These results demonstrate that the proposed framework is robust, scalable, and experimentally verified for intelligent microgrid control, bridging the gap between simulation-based energy management strategies and real-world implementation with potential applicability to both residential and industrial energy systems.