Optimization And Allocation Of Car Sharing Service Resources Under Uncertain Information

In recent years,many environmental pollution problems and social problems have been brought by the rapid development of industry and the large urban traffic congestion,which significantly reduce the speed of overall socio-economic development and people's living standards.Under the background of sharing economic and supply-side reform,vehicle sharing service becomes an innovative service model to solve these problems with the continuous development of network technology and the popularization of mobile payment.In the process of vehicle sharing service operation,customer behavior,environment and other factors contain a lot of uncertainties,such as order demands,and rental time.On the one hand,these uncertain factors make the normal operation inefficient,on the other hand,they also lead to imbalance in the rental stations and other issues,and increase the operation costs of vehicle sharing service operators,and thereby they reduce many social benefits.In addition,operators need to ensure that the overall service continues to be stable,which requires the optimization of its resource and configuration to provide more robust decision results.Robust optimization and distributionally robust optimization can be used to provide robust decision results with uncertain information(such as order quantities,net number of returned cars,etc.)to further optimize the allocation of vehicle sharing service resources and make it perform better profitability and social benefits.Based on the uncertain information in vehicle sharing service,this dissertation uses many methods of dealing with uncertainty to divide vehicle sharing service into site location,initial vehicle configuration,and vehicle reposition based on operation management three stages "strategic management","tactical management" and "operation management".And we research the optimization of resources and configuration in the three targeted stages.The main research contents are as follows:(1)Based on the probabilistic coverage location model and the adjustable uncertainty set,a robust probability coverage location model is established for the preuncertain potential lease demand.Firstly,the customer adoption rate is introduced to reflect the heterogeneous travel utility of customers.Combined with the probability coverage location model,the nonlinear model is transformed into the linear model by Bonferroni inequality under the worst coverage condition.Secondly,when the number of potential demand is known in early stage,it is assumed that the range of potential demand is known,and we introduce the adjustable parameters to adjust the variable range,reflect the partial changed demand and adjust the degree and the number of disturbances for further optimizing site location decision.The results show that the location and configuration of the charging pile at the stations under the uncertain information.We get that the actual demand disturbance and the heterogeneous utility dispersion can increase the operation cost.And the fast charging pile with high service efficiency can effectively improve the operational efficiency and reduce operation costs.(2)In order to meet the requirements of the uncertain orders in early stage,we consider three conditions,certain demand,the random demand,and uncertain demand under partial moment information,and establish the certainty model,stochastic model and distributionally robust model.Firstly,it is assumed that the model is configured under the certain demand and it is calculated by setting the variable parameter to simulate the actual lack of vehicle.Secondly,by setting the demand distribution into two points distribution with the smallest amount of information,we get the actual lack of vehicles in the worst case to configure vehicles with conservative form.Finally,we assume that the first-order moment and second-order moment of demand information is known,the distribution constraint sets are transformed into the positive definite programming by the duality theory,and the distributionally robust vehicle configuration model is established.The results of the study demonstrate the amounts of vehicles in the three cases based on the results of location.It shows that the distributionally robust vehicle configuration scheme proposed in this dissertation can save the configuration cost for the operators to meet the demand of partial information with fewer vehicles.(3)Design the distributionally robust vehicles reposition model under partial statistical information for the unbalanced problems in the uncertain traffic volume and sharing service at the operational level.Firstly,the net traffic flow reflects the uncertainty of the demand with vehicles,and it is assumed that the net traffic flow has part of the statistical information,we build the opportunity set constraints of available vehicles and idle charging piles under the worst case.Combined with the second moment of uncertainty factors' distribution sets,the method of transforming chance constraints based on the probability is proposed,and the semi-definite programming is transformed into a second order cone optimization for further optimizing vehicle reposition.The results of the numerical study indicate the number of vehicle repositioned between the stations and find that the reposition cost and the total traffic flows increase because of service level promotion,and the different demand information in different periods leads to service inconsistency.In this dissertation,we focus on the key issues in car sharing service,consider the uncertain demand information in service operation,combine the three stages of operation management,and optimize the site position and vehicle layout,and vehicle reposition.This research enriches the field of pattern management in sharing service,especially for car sharing service.It is of great practical significance to determine the practical operation factors.The robust optimization methods and customer behavior handling methods proposed in this dissertation reduce operating costs and enhance customers' satisfaction,further enhance the quality of service and provide theoretical support to car sharing service,and make it play its social utility of easing traffic congestion and air pollution as well.