EV charging scheduling with renewable energy-powered Green Charging Stations: A Multi-Agent Deep Deterministic Policy Gradient approach

Abstract

With the fast growth in Electric Vehicles (EVs), the demand for electric energy of Internet of Vehicles (IoVs) has increased intensely, resulting in severe issues such as energy efficiency, energy shortage, and carbon emissions. Renewable energy-powered Green Charging Station (GCS) has the potential to solve the above issues, whereas multiple challenges exist in utilizing renewable energy to power IoVs efficiently, such as the uncertainty and fluctuations of both renewable energy and EV charging demand, as well as the inherent influence of user’s preference and behavior on the charging performance. In this paper, we aim to solve these challenges in a typical scenario where charging stations are interchangeably powered by renewable energy sources and grid. A Generative Adversarial Network (GAN)-based forecasting algorithm is designed to predict the renewable energy generation process accurately. Furthermore, a decentralized charging scheduling method based on Multi-Agent Deep Deterministic Policy Gradient (MADDPG) is developed, which constructs the charging scheduling problem as a Markov Decision Process (MDP) and effectively addresses the problem of matching EVs and GCSs, planning EV’s traveling route and selecting charging mode simultaneously. Extensive simulation results demonstrate the effectiveness and superiority of the proposed method in terms of users’ satisfaction, system cost and total time.