Akademik Veri Yönetim Sistemi
Electrica, cilt.23, sa.2, ss.202-211, 2023 (Hakemli Dergi)
Microgrids emerge as a structure that gains more importance day by day with the reduction of energy loss rate, the efficient use of renewable energy sources, the possibility of autonomous operation with energy storage systems, and the profitability it offers. Furthermore, this structure, which helps to reduce the carbon footprint, will become undeniably critical to use in the near future with the nanogrid and smart grid. An innovative dynamic energy management system will make these advantages offered by the microgrid more accessible while facilitating the integration and effective contribution of electric vehicles. On the other hand, thanks to promising and useful developments and algorithms in machine learning and deep learning, artificial intelligence (AI)-based control methods and applications are constantly increasing. Accordingly, the concept of reinforcement learning (RL) offers an unconventional perspective on the control of systems. This study, which is the last step of creating an AI-based energy management system, presents a graphical interface design in light of all these requirements and developments. In this study, the deep RL agent used in determining management actions, together with the prediction models created to make the necessary predictions, are gathered under a single roof.