物联感知环境下电动汽车充电等待时间分布的预测方法

Prediction Method of Charging Waiting Time Distribution for Electric Vehicles in Internet of Things Perception Environment

已有的充电站等待时间研究往往忽略不同车辆充电需求的时变差异性,为全面描述充电等待时间的时变规律,文中提出了物联感知环境下电动汽车充电等待时间分布的预测方法。首先,确定充电站到达车辆数分布和充电时长分布,采用M/G/n排队模型模拟充电站排队等待系统,求解电动汽车充电等待时间分布函数。其次,建立用户选择充电站效用函数,利用Multi-logit模型对各个充电站内的充电需求进行预测。再者,根据充电需求预测值和网联充电站实时数据更新平均到达率、服务时间变异系数,提出电动汽车充电等待时间短时分布的预测算法。最后,以某城市某区域为例,验证了该分布的准确性,并通过路径选择分析了充电等待时间分布和可靠性对充电站的影响,为减少充电站排队拥堵和均衡电网负荷提供决策依据。

In existing studies on waiting time of charging stations, the time-varying differences in charging requirements for different vehicles are often neglected. A prediction method of charging waiting time distribution for electric vehicles(EVs) in Internet of Things(IOT) perception environment is proposed to comprehensively explore the time-varying law of charging waiting time. Firstly, on the basis of determining the quantity distribution of vehicles arriving in the charging stations and the charging duration distribution, the M/G/n queuing model is used to simulate the queuing system in charging stations and solve the distribution function of charging waiting time for EVs. Secondly, the utility function for user to select charging stations is established and the Multi-logit model is used to predict the charging demand in each charging station. Furthermore, the average arrival rate, service time and variation coefficient are updated according to the predicted value of charging demand and the real-time data of network-connected charging stations, and the short-time prediction of EV charging waiting time is realized with the distribution function of charging waiting time. Finally, taking a certain area of a city as an example, the accuracy of this distribution is verified, and the influence of charging waiting time distribution and reliability on charging stations is analyzed, which provides a decision basis for reducing queuing congestion and balancing power system load.