Real-Time Bidding Strategies from Micro-Grids Using Reinforcement Learning

We address the problem faced by the operator of a microgrid participating in a continuous real-time market. The microgrid consists of distributed generation, flexible loads and a storage device. The goal of the microgrid operator is the maximization of the profits over the entire trading horizon, while taking into account operational constraints. The variability of the Renewable Energy Sources (RES) is considered and the energy trading is modeled as a Markov Decision Process. The problem is solved using reinforcement learning (RL). The resulting optimal real time bidding strategy of a microgrid is discussed.