Intelligent charging and discharging scheduling in multistorey car park with wireless electric vehicle network

Abstract

Multistorey car park outfitted with wireless charging infrastructures can offer space-efficient and charging-convenient solutions for electric vehicles (EVs) in urban areas. Serving as energy storages, numerous EVs parked for long periods hold substantial potential to alleviate transformer load stresses and yield considerable profits for drivers. To effectively realize dual benefits, this paper investigates the EV (dis)charging scheduling problem, aiming at maximizing the driver’s profit while guaranteeing sufficient battery’s state of charge upon departure, and avoiding transformer overloading. A multi-agent reinforcement learning method is utilized based on an actor-critic framework. Also, a revised centralized training and decentralized execution is designed based on a simple statistic global observation, which is able to address the time-variable number of agents problem. The computational simulation is conducted to verify the feasibility of the proposed algorithm.