Handling Computation Hardness and Time Complexity Issue of Battery Energy Storage Scheduling in Microgrids by Deep Reinforcement Learning

With the development of microgrids (MGs), an energy management system (EMS) is required to ensure stable and economically efficient operation MG system. In this paper, intelligent EMS proposed by exploiting deep reinforcement learning (DRL) technique. DRL employed as effective method for handling computation hardness optimal scheduling charge/discharge battery storage in EMS. Since decision depends on its state charge given from consecutive time steps, it demands a full-time horizon obtain optimum solution. This, however, increases complexity turns into NP-hard problem. By considering system’s charging/discharging power control variable, agent trained investigate best both deterministic stochastic weather scenarios. The efficiency strategy suggested study minimizing cost purchasing also shown quantitative perspective through programming verification comparison with results mixed integer heuristic genetic algorithm (GA).