Research On Travel Patterns And Rebalancing Optimization Problem In Free-floating Bike Sharing System

In recent years,with the rapid development of mobile payments,free-floating bike sharing(FFBS)has sprung up in China since March 2016.It plays an important role in urban transportation system due to the convenience of picking up and returning bikes,the convenience of payment and high accessibility.However,faced with this new thing,government and businesses may encounter difficulties due to the lack of corresponding management experience,such as the difficulty in renting FFBSs during peak hour,unequal supply and demand in some regions,massive disused FFBSs,and unresponsive redistribution,which seriously restricts the level of service and operating efficiency in FFBS system.In order to solve these problems,the research on the analysis of different travel patterns and the rebalancing problem with considering the collection of malfunctioning bikes in FFBS system were carried out.They have certain theoretical and practical significances for improving the user satisfaction and achieving the balance between supply and demand.Firstly,the similarities and differences between FFBS system and public bicycle sharing(PBS)system were identified from the system perspective.Then,the characteristics and influential factors of FFBS system were analyzed,and three different travel patterns were summarized.On this basis,the uniqueness factors in the reposition problem and the method to distinguish malfunctioning bikes were proposed.Secondly,the travel characteristics of these three travel patterns: Origin to Destination Pattern(ODP),Travel Cycle Pattern(TCP)and Transfer Pattern(TP)were investigated based on the questionnaire survey data of FFBS users in Nanjing.A multinomial logit(MNL)model was used to quantify the impact of various variables on travel pattern choice,and corresponding intervention policies and strategies regarding ticket pricing,station introduction pattern,and malfunctioning bikes handling were proposed.Thirdly,an integer linear programming model for the rebalancing problem with multiple depots,heterogeneous rebalancing vehicles and multiple visits in FFBS system with considering the collection of malfunctioning bikes under multi-objective condition was established.The model was solved accurately by CPLEX.In the case of the large scale of bike stations,an efficient greedy-genetic algorithm was designed,in which the selection of adjacent station sets and the repair strategy of solutions could help improve the efficiency of the algorithm.Finally,Share-A-Bull FFBS system in Tampa,South Florida,was used to verify the validity of the greedy-genetic algorithm.Sensitivity analysis of the number of rental stations and the capacity of rebalancing vehicles were carried out.Furthermore,the efficiency improvement in different proportions of malfunctioning bikes was quantified by using the data of Divvy PBS system in Chicago.The results showed that the rebalancing problem with considering the collection of malfunctioning bikes had obvious advantages over the separate rebalancing problem of normal bikes and malfunctioning bikes.