Artificial neural network based hierarchical intelligent control framework for a residential microgrid

As Saudi Arabia accelerates its transition to intelligent, sustainable energy systems under Vision 2030, advanced control of renewable microgrids becomes critical. This study proposes an artificial neural network-based hierarchical intelligent control framework for a fully renewable hybrid microgrid powering a residential villa in Jeddah, Saudi Arabia. The system integrates a 36 kW solar PV array, 10 kW wind turbine system, and 100 kWh lithium-ion battery to meet a daily load of 177.5 kWh. Neural networks are embedded across all control layers: MLP-based MPPT for solar and wind sources, NARMA-L2 for battery management, and an intelligent EMS for system-wide coordination. Simulations conducted in MATLAB/Simulink, under realistic variable irradiance (0.1-0.8 kW/m²) and wind speeds (3.5-5 m/s) characteristic of Jeddah, quantitatively demonstrate the superior performance of the proposed framework. Specifically, the MLP-based MPPT controllers achieves a power tracking enhancement of up to 12% compared to conventional MPPT methods, significantly improving energy harvesting efficiency. The NARMA-L2 battery management controller exhibits a reduction in DC bus voltage error by approximately 10% and an improvement in recovery time by over 5% when benchmarked against traditional PI control, underscoring its enhanced stability and responsiveness. Furthermore, the intelligent EMS successfully maintained real-time load balancing and optimized battery operation across five distinct dynamic load scenarios, confirming its robustness. The simulation results validate the effectiveness, resilience, and scalability of the proposed neural network-based hierarchical control system, positioning it as a promising solution for smart residential microgrids in challenging coastal environments.