AI-Enabled DC Bus Control for Hybrid Residential Energy Systems

Purpose: This study aims to enhance DC bus voltage regulation and battery operation reliability in a large-scale hybrid residential microgrid through an intelligent predictive control approach. Methodology: A hierarchical control framework is proposed in which a Nonlinear Autoregressive Moving Average with Exogenous Inputs (NARMA-L2) neural network is implemented as a secondary predictive controller for DC bus voltage regulation and battery management. The controller is designed to learn the inverse dynamics of the DC bus–battery system and anticipate voltage disturbances caused by renewable variability and load changes. The framework is applied to a redesigned hybrid residential microgrid supplying a high-consumption villa in Jeddah, Saudi Arabia, comprising a 40 kW photovoltaic array, a 15 kW wind turbine, and a 100 kWh lithium-ion Battery Energy Storage System (BESS), serving a daily energy demand of 177.5 kWh. Performance evaluation is conducted using MATLAB/Simulink under realistic environmental and load profiles representative of Jeddah conditions. Findings: Simulation results demonstrate that the proposed NARMA-L2-based control strategy significantly improves DC bus voltage stability compared to a conventional PI controller. The DC bus voltage Root Mean Square Error (RMSE) is reduced by approximately 68% (from 3.15 V to 1.01 V), and voltage recovery time is improved by over 7%. In addition, the enhanced generation capacity and predictive control framework increase renewable energy utilization by about 12%, while maintaining battery State-of-Charge (SOC) within safe operating limits and ensuring stable power balance. Unique Contribution to Theory, Practice, and Policy: This study provides practical evidence of the effectiveness of neural network-based predictive control for voltage stabilization in large-scale residential microgrids. The proposed framework bridges the gap between conventional rule-based controllers and intelligent data-driven control strategies, offering a scalable solution for high-demand residential applications. From a policy perspective, the results support the deployment of advanced control technologies as a key enabler for resilient residential microgrids aligned with Saudi Vision 2030 sustainability objectives.