AI-driven real-time control architecture for hybrid battery and supercapacitor systems to enhance frequency regulation in high renewable microgrids

A The growing penetration of photovoltaic (PV) and wind generation in microgrids poses problems of fast-changing power and low inertia, which pose challenges for frequency regulation. Although traditional battery energy storage systems have the capacity for providing continuous energy supply, they might be subjected to increased stress due to the charging and discharging activities in response to high-frequency disturbances. The current paper aims at presenting an artificial intelligence-based real-time control architecture for the coordinated operation of a battery-supercapacitor hybrid energy storage system in a microgrid characterized by a high penetration of renewable energy sources. The developed architecture is able to make the necessary adjustments based on the values of frequency deviation, rate of change of frequency, renewable generation, load demands, battery state of charge, and supercapacitor voltage. The proposed scheme was simulated using MATLAB/Simulink under the conditions of moderately varying renewable sources, sudden load disturbances, highly intermittent renewables, and a combination of renewable and load disturbances. The maximum frequency deviation was lowered from 0.32 Hz to 0.18 Hz (−43.8%), and the frequency settling time decreased from 4.1 s to 2.6 s (−36.5%). The stress on the battery caused by high current during charging/discharging decreased by 31.2%, whereas the estimated increase in the battery lifetime was 27.6%. The energy utilization efficiency rose from 88.2% to 95.4%, and the battery state-of-charge (SOC) remained within the 35–85% operational limits. It is shown that the cooperation of storage systems can help increase the frequency stability and reduce the battery stress in a high-renewable microgrid. Future research will include hardware-in-the-loop (HIL) tests.