| Electric vehicles are gaining widespread attention in society as a low-carbon,environmentally friendly and economical means of transportation.However,with the speedy growth of the number of electric vehicle owners,there is a serious gap between the provision and needs of charging services,and the layout and allocation of charging infrastructure significantly affects the promotion and popularization of electric vehicles.Therefore,finding practical ways to rationalize the layout and capacity allocation of charging stations in an efficient manner is crucial.Faced with the problems that the key parameters of charging load prediction researches are not detailed enough,the accessibility of charging stations is less studied,and the link between charging station planning and electric vehicle purchasing behaviors is not close enough,this paper thoroughly investigates the problems of charging station accessibility and locating and sizing optimization based on accurate prediction of charging load spatial and temporal distribution.The main research works of this paper are as follows.(1)To address the problem that the key parameters of charging load prediction researches are not detailed enough,a method for predicting the spatial and temporal distribution of charging loads that adequately describes the travel and charging behaviors of electric vehicles is investigated.Firstly,the temporal characters of electric vehicle user travel chain and battery energy information are carefully simulated and portrayed,and the real road network of the research area is extracted based on Openstreet Map.Then the speed-flow road resistance model is introduced to simulate the relationship between the driving speed and road traffic flow of electric vehicles in each road section of the traffic road network.And then,the OD matrix is inverse extrapolated based on Trans CAD,the Floyd algorithm is adopted to simulate the driving paths of electric vehicles between the origin and destination,and the process of electric vehicle charging load prediction based on Monte Carlo method considering spatial and temporal distribution is proposed.Finally,the spatial and temporal distribution of charging load in the research area is predicted and analyzed,and the results show that the careful portrayal of travel and charging behaviors greatly enhance the correlation between charging loads and travel behaviors for different types of electric vehicles.(2)In response to the lack of research on the accessibility of charging stations based on real data and integrated time-variant supply,demand,and mobility data,a charging station accessibility analysis framework based on G2 SFCA is proposed.The spatial distribution of24-hour charging station accessibility is measured based on the input of time-related variables,and the accessibility is temporally clustered by agglomerate hierarchical clustering algorithm to categorize the 24-hour accessibility into five periods.Then the correlation between dynamic accessibility measurements and temporal clustering measurements is verified using Pearson correlation coefficients.The results show that the proposed framework can accurately describe the spatial and temporal dynamics of charging station accessibility and the spatial unevenness of accessibility during peak travel periods in the research area,which is helpful for the research of locating and sizing optimization of charging stations.(3)To address the problem that charging station planning is not closely linked with electric vehicle purchasing behaviors,a two-level optimization model based on IDCM for charging station locating and sizing is proposed.The upper-level operator decision objective is to maximize the purchase rate of electric vehicles,and the lower-level is optimized with the objective of maximizing the user option utility.To overcome the problem that the non-linearity and non-convexity of IDCM probability formulas limit the solution of the optimization model,the error term of the utility function is expressed as a linear combination of the random vectors from IID normal distribution and IID Gumbel distribution,and the lower-level model is described as the maximal coverage location problem(MCLP),and finally the two-level optimization model is converted into an integer linear programming(ILP)problem and solved by an exact algorithm.The results show that the spatial distribution of electric vehicles purchased is significantly affected by the locating and sizing optimization based on the consideration of user purchase behaviors,and the electric vehicle purchase rate increases by 33.85% with optimal operator investment.(4)To improve the efficiency of solving the charging station locating and sizing optimization model,three heuristic algorithms are applied and the results are compared and analyzed.Firstly,the processes of rolling horizon,greedy algorithm,and greedy randomized adaptive search procedures(GRASP)for solving the charging station locating and sizing optimization model are proposed.Then the greedy function,greedy parameters and RCL length,which are the main parameters influencing the performance of the GRASP algorithm in the initial solution construction phase,are set,and three neighborhood structures are designed and the neighborhood search method is determined in the local search phase.To enhance the solving efficiency of the GRASP algorithm,a new filtering stage is added between two stages to filter out invalid solutions as early as possible.Finally,three heuristic algorithms are used to solve the case,and a larger-scale case is presented to further validate the applicability of the GRASP for solving the charging station locating and sizing optimization model.The experimental results show that the GRASP is more advantageous than the mathematical planning exact algorithm,and is considered to be the optimal solution algorithm among all algorithms in larger scale cases,and it is very applicable to solving the charging station locating and sizing optimization model proposed in this paper. |
PDF Full Text Request