公共自行車網絡新設站點的分析方法

Abstract

近年來，公共自行車系統在全世界大小城市都非常盛行，在多數系統內的站點數越來越多，逐漸趨於飽和的情況下，仍然有無車可借或無位可還的需求不平衡問題，但如果再持續新建越來越多的站點卻會對財務上造成很大的負擔，調度上的困難也隨之增加，故而本研究針對新設站點的問題進行了分析和研究，考量附近站點的需求和位置，提出一套方法以求出最佳新站點位置，希望每一個新站點都能對減少需求不平衡的問題起到很好的效果。 本研究首先建構時空模型(Time Space Model)來描述系統中自行車數量隨著時間的變化，而後分別針對新設單個站點和多個站點設計一套方法來求解最佳位置。對單個站點建構一個非線性最佳化模型並用Lingo進行求解，在限制新站點可以服務到的距離的情況下，希望最大化其支援量，也就是提供車輛或車位補足附近缺車或缺位站點的需求；而對多個站點則透過K平均分群(K-means clustering)將需求車輛或車位的站點進行分群，之後用質量中心(Center of mass)求解新設站點的位置，目標是最大化多個新站點支援量的總和，同時最大化這些新站點涵蓋的範圍。後面用台北市YouBike系統作為例子進行分析，求解出最佳解位置並和現有的部分站點做比較，證實求解出的結果較佳。研究成果可以提供給各公共自行車系統的管理者做為未來新增設站點的參考。

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.