Maintaining flexibility in smart grid consumption through deep learning and deep reinforcement learning

The smart grid concept is key to the energy revolution that has been taking place in recent years. Smart Grids have present research since their emergence. However, scarcity of data from different sources, hardware power, or co-simulation environments hindered development. With advances multi-agent-based systems, possibility simulating behavior combining real building consumption, and simulated data, storage batteries vehicle charging points, opened up. This development resulted much published using both physical data. All these investigations show main problem machine learning algorithms do not fully match behavior, it complex use them replicate actions be performed. paper aims combine approach prediction with state-of-the-art techniques, such as deep reinforcement learning, simulate unknown critical system scenarios. A very important element grids maintaining consumption within specific ranges (flexibility). For this purpose, we made Tensorflow libraries predict select optimal performed our system. developed platform flexible enough include new technologies batteries, electric vehicles, etc., oriented real-time operation, being applied an on-going project European ebalance-plus project.1