A Semi-Decentralized Real-Time Charging Scheduling Scheme for Large EV Parking Lots Considering Uncertain EV Arrival and Departure

Developing large commercial electric vehicle (EV) parking lots to support the rapid EV adoption arouses interest in optimizing their real-time charging schedules with enhanced economic efficiency. This problem has been studied in literature via fully centralized or decentralized schemes, i.e., EV charging schedules are solely determined by the parking lot central operator or individual chargers, confronting the dilemma of scalability and parking-lot-wise economic optimality. This paper studies a semi-decentralized real-time charging scheduling scheme, in which the central operator and individual chargers collaborate to achieve optimal EV charging schedules. Specifically, the central operator uses a chance-constrained model to estimate aggregate charging energy needs in a rolling process at a coarse time granularity, while considering uncertainties of aggregate arrival and departure EV charging demands via a Gaussian mixture model; with the estimated aggregate charging energy, the central operator further calculates charging energy references of individual chargers regarding their distinct charging urgency and discharging availability; each charger finally determines the actual charging power by leveraging the charging dynamics, EV departure uncertainty scenarios, and charging energy reference at a fine time granularity. The economics and efficiency of the proposed scheme are evaluated by comparing it to various forms of fully centralized schemes via numerical simulations. Simulation results demonstrate that the proposed scheme, with proper settings on the charging urgency factor, time granularity, and discount factor, significantly enhances efficiency in the minute-wise charging scheduling of large-scale EVs at the slight cost of compromised revenue.