Dynamic Optimization Model And Algorithm Design Of One-Way Electric Carsharing Relocation With Considering User Flexibility

The sharing economy has attracted wide attention since it was proposed,and the car-sharing model has become more and more popular with the development of the Internet economy.Car-sharing modes mainly include one-way car-sharing mode,round-way car-sharing mode and free-float mode.Among them,the one-way car sharing mode is the most popular because it is convenient for users and more feasible than the free-float mode.At present,with the development of electric vehicles,electric vehicles are also welcomed by people because of their environmental protection and emission reduction.The one-way electric car-sharing system has become one of the most widely used carsharing service systems in recent years.However,there are still many problems with one-way electric car-sharing.For example,the uncertainty of demand,the imbalance of vehicle inventory,the shortage of parking spaces,the charging of electric car and the problem of users refusing orders before departure all pose challenges to the operation of car-sharing platforms.Many car-sharing platforms,similar to Go Fun,offer incentives for users who are willing to accept the platform’s flexibility to change their departure and destination stations.Such platforms employ a policy of reserving parking spaces.Through such strategies,the platform can minimize the problem of vehicle inventory imbalance and insufficient parking spaces.At the same time,the shared car platform also uses employees to relocate vehicles to satisfy users and maximize the platform’s profits.However,because such problems are generally large in scale,it is difficult to solve them with exact algorithms,and platforms generally use greedy algorithms for deployment,which may result in unbalanced inventory and fail to meet user needs well.In view of this situation and problems,the research are as follows:A rolling horizontal decision framework is proposed for managing the operation of a one-way electric vehicle sharing system in real time.The dynamic decision framework integrates the operator’s vehicle and staff deployment decisions,the user’s space flexibility,and parking space reservations,while taking EV charging as a constraint.A network flow optimization model is developed for each epoch decision,where any decision that has finally gone off remains unchanged,and decisions that haven’t taken off are re-optimized again.A user’s trip corresponds to a path in the network,and the theoretical properties of its optimal solution are described.Therefore,a new path-searching heuristic is developed that iteratively improves a given solution by solving the shortest path problem.The algorithm searches for multiple local optimal solutions within a given time limit and selects the best one as the final solution.The experiments are conducted using data from the Go Fun platform,and compare our heuristic algorithm with the first-come,first-served(FCFS)greedy strategy,and the results show that our heuristic algorithm can significantly improve the system profit compared with the FCFS strategy.In order to deal with the problem of users rejecting orders before departure,A new two-stage model is also be developed to solve the customer regretting problems.And a new heuristic solving algorithm is developed for this model.Data from Go Fun is used as examples to construct numerical experiments.The results show that the algorithm in this paper is significantly better than the greedy algorithm based on first-come,first-served in terms of profit and service level.