Research On Space-time Network Based Electric Vehicle Dispatching Method For Carsharing System

Carsharing is a new kind of transit serving system,which adds the ideas of “sharing economy” and “internet plus” into the traditional transit system.Carsharing system separates the ownership and the right of using a car by the mode of short-time rent.This will increase the utility of vehicle and control the number of vehicles in big cities.The users can access the vehicles by request for a reservation,while the carsharing companies will collect all the user traveling information and determine the corresponding system operation strategies.The number of vehicles in each station are usually unbalanced during the operation cycle.Therefore,the fleet managing staff are dispatched to rebalance the number of vehicles and ensure that the system is running efficiently and economically.Besides,restricted by the vehicle battery technology,electric vehicle(EV)need longer charging time with poor cruising performance.The charging and consuming features shall thus be paid more attention when making the EV dispatching plan.First of all,a staff dispatching model is formulated basing on the space-time modeling method.Comparing to the traditional vehicle routing operation model,the former is apparently more concise and straightforward.To improve the performance of the algorithms when solving the practical problems,the LR algorithm is adopted to decompose the primal problem into a series of sub-problems.Then these subproblems are solved seperately and the feasible solution can be obtained adapting from the relaxed solution,i.e.,the optimized staff dispatching plan.Then,restricted by the battery capacity,it often takes a long time for EVs charging in a station until energy is enough for the next task.Hence,the formulation of the charging and consuming process of EV is one of the important problems.After carefully investing the charging and consuming characters of EVs,the charging and consuming process is formulated based on the numerical fitting method.Then the formulation is added into the space-time-state network and then the optimization of EV dispatching strategies.This formulation can not only capture the practical charging and consuming process of EV but also reduce the complexity of the model.Therefore,the structure of the model is brief enough for ensuring the performance of the LR when dealing with the real-word scale instances.Last but not least,an integrated framework for electric vehicle rebalancing and staff relocation in one-way carsharing systems is proposed.To further improve the reliability and practicability of the algorithm when solving the practical problem of the integrated formulation,the LR is improved by adopting the branch-and-bound algorithm(LR-BB).To verify the performance of LR-BB,a real-world case study based on the operation data of Seattle,WA is proposed for the numerical experiments.It can be seen from the results that LR-BB is more efficient and accurate enough for dealing with the real-word cases.Besides,a series of managerial insights are drawn from the lay-out plan and operation strategies of staff and EVs from the near optimal results and the sensitive analysis.They provide the managers of carsharing systems with the valuable suggestions to improve the efficiency of carsharing system with less system cost.In this study,a space-time network based modeling framework are proposed for the carsharing dispatching problem.model of staff dispatching proble m,EV dispatching problem and staff-EV coordinated optimization are formulated and solved to obtain the near optimal solution efficiently and accuratly,which provide the scientific basis for improving the service quality and operation efficiency.