A Self-organizing Equilibrium-Oriented Relocation Optimization Algorithm Of Car-sharing

Recently,China's car-sharing industry has developed rapidly.With the introduction of relevant strategy documents concerning the preferential promotion of electric vehicles in the public sector,many electric vehicles have been invested into car-sharing operations.The car-sharing projects have made significant effect in large and medium-sized cities such as Beijing,Hangzhou,Shanghai.However,due to the restriction of charging and maintenance of electric vehicles and the fact that the electric vehicles needs to be picked up and returned at designated rental stations,users need to take into consideration vehicle distribution and vehicle status at surrounding stations when leasing electric vehicles,which indicate that the vehicle distribution between stations directly affects potential users' demand for the vehicles.Since there exists imbalance of demand distribution,that is,the distribution of vehicles after daily operation does not match the spatial distribution of demand the next day at many stations,which leads to the dilemma of "no car to borrow,no car to return" for the same demand in the next day.So the vehicle relocation problem between vehicles needs to be solved for the sustainability of car-sharing operation.Based on the investigation of the operation process and mode of car-sharing,this paper gives full consideration to the factors influencing users'behavior of car-sharing station selection,and proposes a discount-pricing strategy to guide users to pick up and return vehicles at designated stations to optimize the vehicle relocation between stations.Firstly,through the design of SP questionnaire and the statistics of questionnaire data,the factors affecting users'behavior of car-sharing station selection are analyzed.A MNL model of users' behavior of station selection is established according to three influencing factors:vehicle driving range,vehicle pickup walking distance and travel cost.Based on some premise and hypothesis,a car-sharing relocation optimization model based on self-organizing equilibrium is established.On the premise that the distribution of vehicles at each station is consistent in the operation status of the morning and evening,a relocation scheme combining induction of user self-organizing participation in relocation and employee relocation is proposed where gradient discount strategy is designed to induce users to pick up and return cars at designated stations,and relocation cost is calculated and analyzed and a scheme aiming at minimizing both the preferential cost of inducing users to participate in relocation and the relocation cost of employees is proposed.Then,the relocation task set and relocation path are solved and optimized.Firstly,the transportation problem is constructed according to the difference between the number of vehicles in each station and the number of original vehicles after the end of daily operation,and the relocation task set is worked out with the aim of minimizing the total relocation distance.Then,the minimum cost insertion method is used to solve the initial relocation path,and the tabu search algorithm is designed to optimize the relocation path.Four kinds of neighborhood structure are proposed according to the path form and characteristics,and the relevant parameters in the algorithm process are set.Finally,part of Haidian district of Beijing is selected as the research scope,and an example is designed to analyze the results of the model algorithm.It is concluded from the analysis that when the preferential discount gradient is 0.9-0.8 and the preferential strategy starts at 18:00 PM,the optimal cost of manual relocation without preferential strategy and with preferential strategy is reduced by 1.9%and 15.5%respectively compared with the initial cost of manual relocation,and the total cost of the optimized schedule under preferential strategies is reduced by 4.6%on average compared with that without preferential strategies,which indicates that preferential strategies can reduce the total relocation cost under certain circumstances.Then the optimization results of model algorithm for different incentive strategy start time,different gradient discount strategy and different vehicle launch scale are compared and the conclusion is as follows:the total relocation cost is the lowest when the preferential strategy start time is moderate;lower gradient discount has the best effect on user participation in relocation tasks and the total relocation cost;the discount strategy can minimize the total relocation cost when the vehicle launch scale is coordinated with the demand distribution and the supply is less than the demand or the supply and demand are nearly balanced.