Dynamic energy scheduling and routing of multiple electric vehicles using deep reinforcement learning

• Published in: Energy (Oxford) Vol. 244; p. 122626
• Main Authors: Alqahtani, Mohammed, Hu, Mengqi
• Format: Journal Article
• Language: English
• Published: Oxford Elsevier Ltd 01.04.2022; Elsevier BV
• Subjects: Algorithms; Consumers; Deep learning; Deep reinforcement learning; Electric vehicle; Electric vehicles; Energy; Energy costs; Energy demand; Energy resources; Energy scheduling; Energy sources; Energy storage; Genetic algorithms; Global economy; Heuristic methods; Learning; Machine learning; Mobile energy network; New technology; Particle swarm optimization; Photovoltaic cells; Power consumption; Power supply; Reinforcement; Route planning; Security; Vehicle routing; Weather; Energy scheduling; Deep reinforcement learning; Electric vehicle; Mobile energy network; Vehicle routing
• ISSN: 0360-5442; 1873-6785
• DOI: 10.1016/j.energy.2021.122626

The demand on energy is uncertain and subject to change with time due to several factors including the emergence of new technology, entertainment, divergence of people's consumption habits, changing weather conditions, etc. Moreover, increases in energy demand are growing every day due to increases in world's population and growth of global economy, which substantially increase the chances of disruptions in power supply. This makes the security of power supply a more challenging task especially during seasons (e.g. summer and winter). This paper proposes a reinforcement learning model to address the uncertainties in power supply and demand by dispatching a set of electric vehicles to supply energy to different consumers at different locations. An electric vehicle is mounted with various energy resources (e.g., PV panel, energy storage) that share power generation units and storages among different consumers to power their premises to reduce energy costs. The performance of the reinforcement learning model is assessed under different configurations of consumers and electric vehicles, and compared to the results from CPLEX and three heuristic algorithms. The simulation results demonstrate that the reinforcement learning algorithm can reduce energy costs up to 22.05%, 22.57%, and 19.33% compared to the genetic algorithm, particle swarm optimization, and artificial fish swarm algorithm results, respectively. •An integrated routing and energy scheduling under uncertainty model is developed.•The model enables energy sharing among multiple EVs at spatial and temporal scales.•The proposed RL algorithm is more computational efficient than heuristic algorithms.•The proposed RL algorithm yield better quality results than heuristic algorithms.•The proposed algorithm has potential to be used for real-time decision making.