Research On Electric Vehicle Fast Charging Demand Forecast And Charging Station Planning Based On Data-driven

As a necessary guarantee for electric vehicles(EV),reasonable planning of charging facilities is of great significance to the development of EV.The spatiotemporal distribution characteristics of EV charging demands is the important basis for the location and capacity determination of charging stations.Due to the strong randomness of EV users’ travelling,driving and charging behaviors,it is difficult to predict the spatiotemporal distribution of EV charging demands.In recent years,the accumulation of research data related to electric vehicles has accelerated,and depicting the random behaviors of EV users by big data has made it possible to predict the charging demands of EV.Previous research has not yet effectively used the " electric vehicle-traffic network-charging station" data.Therefore,the travel characteristics and charging characteristics of EV users,dynamic traffic network,and actual operation data information are organically combined in this thesis,and a forecasting method for fast charging demands of EV based on data-driven is proposed.Oriented by charging demands,the optimal planning model of charging stations is established,which is verified by the simulation.First,this thesis obtained more valuable regeneration spatiotemporal characteristics data of traffic network,including the original-destination(OD)set of urban residents’ taxi demands,the speed set of vehicles on the road network,and the actual driving path set of drivers by in-depth mining and analyzing the trajectory data of online ride-hailing.Based on the point of interest(POI)data,the urban functional areas are identified.Combined with the OD set of taxi demands,the spatiotemporal travel characteristics of urban residents who choose taxi as the travel mode are obtained.Second,an single electric vehicle model is constructed after building the electric private vehicle model based on the travelling and charging characteristics of users and building the electric taxis model based on the travelling characteristics of taxi passengers and the operation characteristics of electric taxis.The dynamic traffic network model is established by graph theory analysis method through integrating the traffic speed set of the road network and the traffic situation data obtained from Auto Navi open platform.Predicting spatiotemporal distribution of the EV fast charging demand through simulating the electric vehicle users travelling behaviors and charging behaviors by "vehicle-traffic network " model,and obtained the conclusions mentioned below.Charging demand probability is positively related to road grades and road network density: EV charging demand number on senior road is 2.67 times as much as the secondary road and when the number of road nodes increases by 5,the number of charging demand number increases by about 180 in a radius of 1000 meters of circular area.Finally,analyzing the behavior characteristics of EV users’ charging queuing based on the actual operation data of public charging stations,the conclusion is drawn that the queuing system in public charging stations conforms to the M/G/k queuing theory,in which the charging service time is subject to multi-Gaussian distribution.Considering the queuing scenario,the optimal charging device configuration method in the charging station is proposed,which aims to minimize the sum of the service cost of the charging stations and the queuing time cost of EV users.Taking public parking lots as charging stations candidate locations,charging stations optimal location planning model is established by integrating operator investment cost and EV charging time cost of users.Use the improved adaptive inertia weight particle swarm optimization algorithm to solve the model,and the effectiveness of charging demand forecasting method and the charging station planning method is verified by simulating in the planned area.