Steady State and Switching Loss Improvements in Hybrid AC-DC Microgrids Using Adaptive Predictive Learning

Hybrid microgrids (HMGs), integrating AC and DC systems, face performance challenges due to system dynamics, external disturbances, and renewable energy demands. Traditional control methods are limited by frequent device switching and static strategies. This research introduces an Adaptive Predictive Learning System (APLS) to reduce switching losses in bidirectional converters within HMGs. Combining Model Predictive Control (MPC) and Deep Reinforcement Learning (DRL), the APLS addresses power flow, demand response, and energy storage, optimizing steady-state performance and handling system constraints. MPC predicts future system behavior to optimize control inputs, while DRL adapts to dynamic environments, improving decision-making and reducing unnecessary switching operations. This approach enhances the stability and efficiency of hybrid microgrids, creating more resilient energy systems.