Stochastic optimization of electric vehicle charging stations

Abstract

With range anxiety becoming the every day problem for Battery Electric Vehicles (BEVs) owners, even more research is being conducted in the field of BEV charging and Charging Stations (CSs) scheduling optimization. In this context our work addresses the problem of BEV charging in an urban environment with no a-priori knowledge of vehicle arrivals. Our system is modeled as a M/G/K queuing system. Two adaptive charging algorithms are proposed, both of them relying on queue stability. The first one charges BEVs up to a percentage of their maximum capacity when charging queues become unstable. The second one when detects instability charges BEVs sufficiently enough to reach their next destination. Both algorithms can be used in combination with an admission control algorithm that does not allow BEVs that do not fulfill certain criteria into the charging stations. The First-Come-First-Serve (FCFS) algorithm is directly compared to our proposed algorithms, with prominent improvement concerning congestion in charging stations and waiting time of electric vehicles. © 2019 IEEE.