Resilient Control for Islanded Hybrid DC Microgrid Integrating Piezoelectric, Solar and Battery

The paper offers a control strategy for an isolated hybrid DC microgrid, which includes solar power (PV), piezoelectric elements, and battery. It describes a model predictive control approach, enhanced with reinforcement learning, to optimize the performance of a hybrid DC microgrid. This approach surpasses the conventional limitations of the standard proportional-integral (PI) controller. The adaptive system is designed to manage output voltage and current regulation, even when the piezoelectric generator operates under varying input voltages, utilizing a Material Testing System (MTS). The experiment was carried out by varying mechanical frequencies to mimic different levels of pressure from different types of traffic and vehicles. As a consequence, the piezoelectric system's input voltage changes rapidly as the pressure changes. The microgrid control system is evaluated under several scenarios, and both its steady-state and dynamic behavior are thoroughly investigated. To assess the resilience of the controller in islanded mode under several load change conditions, a model was developed in MATLAB Simulink. The results demonstrated that the suggested reinforcement learning controller performed superior than the conventional PI controller.