Research On Configuration Optimization Of Station Warehouses And Virtual Stations In Bike-Sharing System

Complex network method has been widely used in many subjects.Transportation is not an exception.Bike-sharing as a newly emerging thing has become a hot research area in transportation and green behavioral science.Bike-sharing is one of the means of green travel.On the one hand,sharing bicycles facilitates people's travel and enriches the way of travel.On the other hand,sharing bicycles improves people's awareness and sense of responsibility for energy saving and emission reduction,which in turn reflects the meaning of green travel.In order to support people using shared bicycles for green travel,it is very important to optimize the configuration of bike-sharing system.This paper makes a systematic study on configuration optimization of station warehouses and virtual stations in bike-sharing system.Chapter 1 and Chapter 2 are the introduction of some research background and basic knowledge.Chapter 3 studies the location of the station warehouses in docked bike-sharing system.We use the data from September 18,2016 to September 24,2016 in Nanjing to construct a bikesharing network.We have improved a community detection method and generalized it to the directed weighted networks with self-loops.We employ it in our bike-sharing network.We compare it with the original method and the directed modularity method.We find that it can overcome the drawback of the original method.Furthermore,we find that our improved method can find the important nodes in the network at the iteration steps of community detection process.We proved this property and apply it to the bike-sharing network to find the important nodes in order to seek out the stations that need to set extra storage points.At last we present a simulation to test our result and find it can significantly decrease the lost demands at the whole day.Chapter 4 studies the virtual stations in free-floating bike-sharing system.The increasing amounts of free-floating bicycle give a serious challenge in parking planning.Dispatching random parking bicycles will increase vehicles workload and offset emission reduction effect of the bicycles.At this chapter we discuss about the virtual stations in a free-floating bike-sharing system.Virtual station is the focus of efficient and environmentally friendly dispatching,which enables the bike-sharing company to achieve cleaner production.We introduce the modifiable areal unit problem and propose a new method called the graded cluster system to mine virtual stations on a multiscale geographic view.The new method can generate more virtual stations more evenly.This not only makes it more convenient for people to use shared bicycles for green travel,but also dense virtual stations can make shared bicycles more visible in people's sight,strengthening people's awareness of green travel.Chapter 5 studies the dynamic electronic fence and fine strategy in a free-floating bike-sharing system.In order to solve the problem of the capacity of the virtual stations and make the virtual stations more suitable for the usage scenarios during the tidal period,we propose the dynamic electronic fence technology.It achieves the purpose of increasing the internal capacity of the station by intelligently expanding the electronic fence outwards,including dynamic electronic fence technology based on area and dynamic electronic fence technology based on ratio.They can intelligently detect obstacles on the expansion path,morning and evening tide periods,and sudden large numbers of borrowing and returning behaviors.We also give a density penalty strategy for electronic fence attacks.In order to regulate the parking behaviours of customers,we also give a fine policy for parking outside the station.It is a strategy that combines a basic fine price plus an additional upper bounded fine related to the distance that exceeds the basic fine distance.The problems studied in this paper are not considered in combination with the rebalancing problem.If it is considered together with the rebalancing problem,the result will be more accurate and the final effect will be more intelligent.