Research On Key Technologies Of Electric Vehicle Charging Stations Deployment Planning

Nowadays,electric vehicles and new energy technology have gone through a rapid development due to rapid consumption of liquid fossil and progressive concern of environment protection.Different from fuel vehicles,electric vehicles have many limitations,such as limited battery capacity,short driving range and long charging time.The wide spread of electric vehicles is serenely impacted by insufficient charging infrastructures,including inadequate chargers,imbalanced distribution and low utilization.When designing a charging scheme,charging service providers have to take charging stations' range and capacity,charging demand,construction cost and other factors into consideration.We propose a charging scheme to deploy chargers based on dwell patterns and queueing theory.Firstly,we analyze charging demand by counting vehicles' arrival rate based on dwell events.To limit charging stations' capacity,we calculate the probability that users is rejected by a charging station according to the queueing theory.We further formulate the problem as a charging rejection probability minimization problem with constraint of construction cost.As the optimization problem is non-linear,we propose heuristic algorithms including greedy method and simulated annealing method to get the optimal charger deployment strategy.We perform extensive experiments using a real taxi trajectory dataset and compare our algorithms with other baselines.We introduce the Shanghai Taxi trajectory dataset and analyze the charging demand based on drivers' dwell patterns.According to simulation results,charging service providers need fewer chargers for each charging station in the early stage to satisfy users' charging demand and they can gradually enlarge the size of charging stations in the later stage.Taking rejection probability as index,we analyze the impact of construction cost,charger numbers,charger's power,size of charging stations.Besides,we analyze the variation of queueing time and residence time with respect to charger numbers.Finally,results show that our heuristics outperform other baselines and the performance of simulated annealing method is even 2% higher than greedy method.