靜態自行車調度與維護問題

STATIC BIKE REPOSITIONING AND MAINTENANCE PROBLEM

本研究提出一個全新、考慮破損自行車維護的靜態自行車調度問題：系統當中的破損自行車不但可以經由調度車收集並運送回倉庫進行整修，而且能以本研究首先提出的現場維修方法來進行維護。接著，本文構建了一道混合整數線性規劃問題，同時決定調度車輛數量、所有自行車站點上正常自行車的裝卸數量、以及所有自行車站點上已收集和實地維修的破損自行車的數量，以最小化總服務時間和所有站點與其目標庫存水平之差的懲罰成本的加權總和。透過臺灣Youbike的實際案例，數據實驗顯示現場維修比單純收集更能減少調度成本和與目標庫存水平的偏差。敏感性分析顯示較短的維修時間、較大的車輛以及較高的偏離目標庫存水平的權重，能達至與目標庫存水平偏離較小和行駛時間較短的調度策略。同時，本研究提出增加了限制服務時間的延伸問題，並透過數值研究顯示更長的服務時間能減少調配車輛的數量和總服務時間。

This paper proposes a new static bike repositioning problem that maintenance operation is simultaneously considered. In this problem, a broken bike at a station can be either collected on the relocation vehicle to be sent back to the depot or repaired on-site to restore its function during repositioning. This on-site repair becomes a new strategy that is examined in this paper and new to the literature. The problem is formulated as a mixed-integer linear programming problem that determines the number of deployed vehicles, the loading and unloading quantities for normal bikes at all stations, and the number of collected and repaired broken bikes at all stations such that the weighted sum of the total service time and the penalties associated with the deviations from the targeted inventory level is minimized. Using the real-world case of Youbike in Taiwan, this paper demonstrates the on-site repair strategy can reduce the repositioning cost and deviations from the target inventory levels more than mere collection strategy. Sensitivity analyses show that the shorter repairing time, larger vehicle, and higher penalty for deviations from the target inventory level can obtain strategies with smaller deviation and shorter travel time. Moreover, a revised problem which includes service time constraint is proposed and the numerical studies show that the increase in allowable service time can reduce the number of deployed vehicles and the total service time.