AI-driven optimal scheduling for hybrid renewable energy systems in medium load industrial zone of Bangladesh

This paper proposes an intelligent, hierarchical energy management framework for hybrid renewable energy systems (HRES) supplying heavily loaded industrial zones. The proposed system integrates photovoltaic (PV), wind turbine, fuel cell, and battery energy storage units into a common DC microgrid architecture, coordinated through multi-layer control. A hybrid converter-level control strategy is developed that combines sliding mode control for the PV interface, model predictive control for the fuel cell converter, and a PSO-based tracker for the wind subsystem to ensure fast and robust voltage regulation. At the supervisory level, a Long Short-Term Memory (LSTM) based predictive scheduler is introduced to generate optimal participation factors for each generation source by exploiting meteorological forecasts, time indices, battery state, and 33 months of real industrial demand data. The proposed framework is validated in MATLAB/Simulink using compressed-time simulations representing more than two years of realistic industrial operation. Results demonstrate reliable demand satisfaction within a normalized range of 100–150 kW, DC bus voltage regulation around 600 V with deviations below 4.5%, and effective state-of-charge management within safe limits. Comparative analysis confirms improved fuel cell utilization and enhanced renewable penetration relative to rule-based dispatch strategies. System resilience is further demonstrated under generator trip contingencies, where coordinated redispatch and storage support maintain load supply continuity. Overall, the proposed AI-enabled hierarchical control architecture offers a practical pathway for efficient and resilient integration of high renewable energy shares in industrial microgrids. • An AI-enabled hierarchical control framework is developed for industrial hybrid renewable energy systems. • An LSTM-based predictive scheduler performs proactive resource dispatch using real industrial load and weather data. • A coordinated hybrid controller set (SMC–MPC–PSO) ensures fast and robust converter-level voltage regulation. • The framework is validated using 33 months of real industrial demand data from a major manufacturing region. • The proposed system demonstrates enhanced demand tracking, DC bus stability, fuel reduction, and contingency resilience.