Ricerc@Sapienza

In this paper we investigate a public Fast Charge (FC) station nanogrid equipped with a Photovoltaic (PV) system and an Energy Storage System (ESS) using second-life Electric Vehicle (EV) batteries. Since the nanogrid is intended for installation in urban areas, it is designed with a very limited connection with the grid to assure peak shaving and encourage PV autoconsumption. To demostrate the effectiveness of this approach to FC stations, an Energy Management System (EMS) is developed to manage the energy demand uncertainty of EVs and the power gap between the grid connection and the FC service. In particular, we propose a machine learning procedure for the automatic synthesis of a suitable (fuzzy) rule-based EMS. Indeed, we posit that a prediction based EMS would result not effective because of the stochastic and intermittent behavior of the FC load, and that a crisp rule-based system, defined by expert knowledge, would be too limited to capture uncertain behavior. The concept is demonstrated in a simulated environment inspired by the “Smart Columbus” project, implementing a mixed deterministic-stochastic process to simulate EV energy demand. In particular, different EV fleets and PV sizes are considered for EMS training, offering insights into the optimal size of PV system and nanogrid system effectiveness. The proposed approach is evaluated by comparing the EMS performance with related optimal benchmark solutions, evaluated by considering known a priori the overall PV and FC demand. The results show that the EMS performance is approximately within 10% of the benchmark optimal value.