Optimization Of Path Choice And Charging Facility Location For Electric Vehicles Under Uncertain Environment

Thanks to continuous attention to the problems of environmental pollution and energy consumption,electric vehicles(EVs)have received widespread attention in recent years,and become a hot topic in the transportation field.Compared with traditional gasoline vehicles,EVs have the advantages of energy saving and environmental protection,low use cost,and good driving experience.However,EVs also face problems such as short driving range,slow and difficult charging during driving,which restrain the wide adoption of EVs.Scientific guidance for EVs' travel and good planning for charging facility layout can reduce travelers' range anxiety and promote the orderly development of EVs.Based on the theory of transportation network flow,the shortest path problem,traffic assignment problem and charging facility layout problem in the stochastic transportation network are systematically studied in this dissertation for EVs' travel.Specifically,the following main contributions are made in this dissertation:(1)Path optimization for EVs' short-distance and medium-distance travel: the constrained reliable shortest path problem.The constrained reliable shortest path model is established in which the stochastic link travel time in transportation network and EVs' driving range are simultaneously considered.It's a generalized shortest path model which can be degraded into the reliable shortest path problem,constrained shortest path problem or shortest path problem according to different parameter setting.The pre-processing and network reduction method is first introduced to reduce the scale of problem.And then the Lagrangian relaxation algorithm and multi-criteria labeling algorithm are designed to solve this problem.Besides,the A* algorithm is integrated into the multi-criteria labeling algorithm to assign priority search rights to node labels closer to the destination node,which further improves the algorithm's computing performance.(2)Path optimization for EVs' long-distance travel: the reliable shortest path problem with charging action.To find the optimal path for the EVs' long-distance inter-city travel,the stochastic link travel time and dwelling time at charging nodes,as well as the EVs' driving range and charging action are considered to be integrated into the shortest path model.Based on different modeling ideas and decision variables,the link-node-state based model and charging-action sub-path based model are established respectively.Corresponding to the two modeling ideas,the multicriteria labeling algorithm and two-stage algorithm are respectively proposed to solve the problem.The application scenarios of two kinds of algorithms are analyzed in the grid network.The multi-criteria labeling algorithm is applicable to the transportation network with highdensity charging nodes and few O-D pairs,while the two-stage algorithm is applicable to the transportation network with low-density charging nodes and many O-D pairs.(3)Traffic flow distribution in the transportation network with the addition of EVs: the reliable traffic network equilibrium problemTaking the uncertainty of traffic demand as the source,the expectation and variance of link travel time and dwelling time at charging nodes are derived analytically,and the reliable travel time of paths is estimated according to the independence hypothesis of travel time and the central limit theorem.Based on the theory of traffic network equilibrium,a multi-user variational inequality model with gasoline vehicles and EVs is established.It is proved that at least one solution exists in the model,and the equivalent expression of Wardrop first-principal is derived.Based on the algorithms for the reliable shortest path problem of EVs' long-distance travel,the basic framework of column generation algorithm is introduced,and the successive average algorithm based on column generation(CG?MSA)is proposed to solve the reliable traffic network equilibrium problem.(4)Layout optimization of charging facility: the discrete network design problem with stochastic traffic demandThe bi-level programming model is established for the layout optimization of charging facility with stochastic traffic demand.In the upper-level problem,the total travel time budget of the transportation network with stochastic traffic demand is analytically derived as the objective function,and the capital investment budget is taken as the constraint condition.In the lower-level problem,a link-based traffic network equilibrium model is established based on the construction of the path travel time budget with subadditivity.The genetic algorithm and CG?MSA algorithm are designed to solve the upper-level and lower-level problems respectively.Finally,a bidirectional ND network is used to analyze the influence of various input parameters on the layout optimization of charging facility.