Charging-Constrained Microgrid Operation Enhancement Based on Uncertain Demand Response

To achieve green and low-carbon development goals, the penetration ratio of renewable energy is gradually increasing in microgrid. Meanwhile, with advancements in energy storage technologies, an increasing number of electric vehicles (EVs) are gradually being integrated into the microgrid, which reduce carbon emissions on both the supply and demand of the microgrid when supplied by clean energy. However, the renewable energy generation and the EV charging load exhibit significant uncertainties. It is of great practical interest to effectively schedule partially-controllable charging demands and fully-controllable loads in the microgrid to improve the self-sufficiency of the microgrid, considering supply and demand uncertainties and the battery life of EVs. This paper addresses the above issues with the following contributions: First, a microgrid optimization scheduling model is proposed based on constrained Markov decision process (CMDP), considering the uncertainties in renewable energy generation, flexible controllable loads, partially-controllable charging demand, and the impact of EV charging cycles. Second, a Lagrangian relaxation-based policy improvement algorithm is proposed to determine the number of EVs to be charged and then a charging sorting heuristic strategy is applied to select the corresponding EVs to charge. Finally, simulation results demonstrate that this method can enhance the operational efficiency of the microgrid.