考慮維修需求之大眾運輸排程之研究

Abstract

由於公車的排程關係到公車的可及性、安全性以及車輛的使用期限，所以公車的排程對於大眾運輸公司來說是相當重要的。一個好的公車排程可以大眾運輸公司的營運更加有效率。然而，車輛的維修也是大眾運輸公司營運中另一個主要的要素，但在現今的營運程序中，營運前公車的預先排程通常未考慮車輛的維修需求。預排的公車排程必須透過調度員的調整才能符合維修需求。 近幾年來，螞蟻族群演算法被應用在許多的實例上，並且螞蟻族群演算法通常比其他演算法有較佳的結果或效率，所以本研究提出一個以螞蟻族群系統(ACS)為基礎之演算法來求解考慮維修需求之大眾運輸排程的問題。 在本研究所提出的演算法中，共有5項重要的課題，分別為(1)網路設計，(2)狀態轉換規則，(3)路徑構建規則，(4)費洛蒙區域更新規則，以及(5)費洛蒙全域更新規則，並且根據這些規則來來求解考慮維修需求之大眾運輸排程的問題。 為證明演算法的實用性，本研究從一家大眾運輸公司取得實際的營運資料，並且產生實際的測試範例來進行測試。經過測試結果證實，本研究所提出的演算法可以解決實務上的問題，並解同時產製公車的營運排班以及維修排班結果。

Because the bus scheduling relates to reliability, safety and vehicle life, the bus schedule is very important to the public transit company. A good bus scheduling can make the operation of the public transit company more efficient. Vehicle maintenance is another major component in public transit operations. But the current operation procedures of bus scheduling is often established before considering the vehicle maintenance requirement. The schedule must be modified by dispatchers to fulfill the requirement. In recent years, the ant colony algorithm are used in many cases. The ant colony algorithm often has better solutions or efficiency better than other algorithms do. So we propose an ant colony system (ACS) based algorithm to solve this problem of public transit scheduling with maintenance requirement. The five major issues embedded in our algorithm are 1.network designing, 2.the state transition rule, 3.the route building rule, 4.the pheromone local updating rule and 5.the pheromone global updating rule. According to these rules, we will solve the problem of public transit scheduling with maintenance requirement. In order to prove this algorithm, we obtain real operation data from a public transit company, and we test these small real samples through generating those testing examples. The testing results confirm that this algorithm can successfully solve the problem of a real operation and meanwhile, it also generates the schedules of the buses operation and maintenance.