Harnessing solar power with adaptive control of PV-enriched microgrids using A3C-driven deep reinforcement learning

Abstract

This research introduces an advanced adaptive control framework utilizing deep reinforcement learning, specifically the Asynchronous Advantage Actor-Critic algorithm, to optimize the operation of photovoltaic-enriched microgrids integrated with solar electric vehicles. The integration of solar electric vehicles within microgrids not only addresses transportation needs and energy sustainability by acting as dynamic energy storage systems. The inherent intermittency of solar power and the dynamic energy demands of solar electric vehicles pose significant operational challenges, requiring robust, flexible control systems capable of real-time optimization. Our framework leverages the Asynchronous Advantage Actor-Critic algorithm for its efficiency in handling high-dimensional state spaces and its capability for rapid, concurrent learning processes, making it well-suited for the dynamic and complex environment of photovoltaic-enriched microgrids. The proposed model innovatively combines solar energy generation with solar electric vehicle energy storage and consumption dynamics, providing a holistic approach to microgrid management that optimizes energy flows, reduces reliance on traditional energy sources, and minimizes environmental impact.