| Summary: | 碩士 === 國立臺灣大學 === 工業工程學研究所 === 105 === Recently, public bike sharing systems are becoming popular around the world. Although the number of stations in many systems constrained by physical space, in many cases, customers tend to demand more stations. However excess stations will cause a great burden on financial and rebalancing difficulties. This paper focuses on the facility location problem of new incremental stations. Considering the requirements of nearby stations, we propose a solution method to find the new station locations.
In this study, we first construct the spatiotemporal model to describe the time-varying state of bike supply in the system, and then design two methods to solve the optimal location for the cases of single station and multiple stations. For the former, we construct a non-linear optimization model. Given the limited distance that a new station can serve, it is desirable to maximize the amount of support which providing an available bike or a parking space for stations nearby. For the latter problem, we use K-means clustering to group the demand stations together and calculate the mass center for each cluster. The goal is to maximize the sum of the support of multiple new stations and the range covered by new stations. Finally, we analyze the YouBike system in Taipei, Taiwan and apply our methods to find optimal solution. Our solution is shown to have better performance than the actual second-phase expansion of the YouBike system.
|
|---|
