市區公車行車系統模擬軟體之設計

市區公車是普遍的大眾運輸交通工具，本研究發展一公車行車系統模擬軟體，可提供公車業者在規劃路線時的參考工具，例如若想要了解當某些營運規劃改變對系統績效的影響(如班距的調整對營運利潤有何影響)，藉由此模擬軟體來模擬系統改變前後之績效指標的變化，可作為決策的參考。也可協助管理者了解目前公車營運情形，進而節省公車業者的成本。我們所發展的模擬軟體是針對一個具有隨機性並且是一個多條路線可重複路段的公車行車系統。因為實際生活中，公車行車系統是隨機的，系統的不確定性導因於：(1)每位乘客到達公車等候站的時間會因人因時而異，(2)每位乘客下車的地點會因人而異，(3)公車的行車時間會隨路況與尖離峰之交通情況而異以及(4)顧客到達站牌候車時會因為排隊長度過長或在排隊隊伍中等候時間過長而離開等候線(稱逃逸現象)。為了使我們的公車行車系統模擬軟體符合實際生活狀況，我們應該考慮這些不確定性因素。本模擬軟體只針對公車行車系統，因此不如交通模擬軟體CORSIM般複雜，模擬的道路只包括市區公車路線行經的道路；模擬的個體（entity）只包括公車與乘客而不包含其他在道路上出現的車輛及行人；模擬輸出只包含與公車績效指標有關的資料（如上車與下車人數、候車人數及時間等），並不包含車流量的相關資料。但相對地，本軟體的優點有二：（一）介面設計簡單且以中文方式呈現可方便公車系統管理者使用，（二）所採用的隨機模擬模式較CORSIM模式的隨機假設廣泛。此模擬軟體採用Borland C++ Builder 5.0程式語言撰寫而成，是一個具有圖形化使用者介面的視窗模擬軟體，軟體具有三項功能：輸入、執行模擬與輸出績效指標。輸入部分包含系統參數(營運資料、路線資料、行車時間、發車時刻表、旅運需求) 和實驗參數(模擬天數及起始亂數種子)。輸入方式除了由使用者從視窗逐一輸入外，還提供使用者直接以開啟舊檔方式讀入輸入參數，以節省輸入時間。在執行模擬部分，通常利用「執行整個模擬」功能鈕來完成整個模擬實驗。也可以利用「一次執行一天」功能鈕每次只模擬一天，顯示模擬當日的細部資料(如事件種類及發生時間、乘客到站時間及目的站、公車到站時間及站間行車時間、逃逸人數、上車及下車人數)。為了偵錯，我們還提供「執行單步」功能鈕，按一次功能鈕就執行下一個事件，用來追蹤每一個事件所執行的結果是否正確。等到模擬實驗結束，會估計此系統的績效指標。在輸出績效指標部分分成系統績效指標(平均每日利潤、平均每人等候時間)、各等候站績效指標(平均候車人數、平均每位乘客等候時間、平均逃逸人數)及各路線績效指標(平均每日上車人數、平均每日發車率、平均每日利潤)等。關鍵詞：隨機模擬模式、多條路線的公車行車系統模擬軟體。

關鍵字

隨機模擬模式；多條路線的公車行車系統模擬軟體

並列摘要

We consider the computer software for simulating the operations of urban-bus systems. This software can be a useful tool for system design. For example, a system manager may want to know how the changes in system design affect the system performance, e.g., how does the change in headway affect the daily profit. With the software, we can simulate the system of interest and observe the system performance before and after the headway change. The simulation results are helpful for making decisions. The simulation model used inside the software is a stochastic system with multiple routes; some routes have overlapped segments. In practice, the bus system is stochastic. The random factors include (i) passenger arrival times, (ii) destinations of passengers aboard, (iii) travel times between bus stops, and (iv) reneging (passengers departing from the queue after a long wait). To mimic the real-world systems, these factors are considered in the simulation model. Our software is designed specifically for bus systems and hence is not as complicate as general-purpose transportation simulation software such as CORSIM. In the simulation model, the routes are bus routes only (not including other street routes), the entity types include bus and customer only (not including other vehicle types or pedestrian), and the system performance measures are all bus related (e.g., customer waiting times, number of passengers on/off the bus, etc.), not including street traffic conditions. On the other hand, this software has two advantages: (i) it is a user-friendly design for bus systems, and (ii) the stochastic assumptions are broader than those used in CORSIM. This software, a graphical user interface, is coded in Borland C++ Builder. Three major features are inputs, execution, and outputs. The inputs are system parameter values (e.g., bus operation data, route data, travel time, time table and passenger demand) and experiment parameter values (e.g., initial random-number seed and number of replications). A user can key in input data from the input windows or read an existing input file if the input data have been keyed in and saved into an input file. In execution, we can click the “run all” to perform the whole simulation experiment. We can also use “run a day” key to simulation the one-day bus operation and observe details (e.g., event list, customer arrival times and customer destinations, bus travel times, number of reneging passengers, etc.). We also provide “step” function button to trace every step of simulation, which is useful for debug. When simulation experiment is over, the performance measures are shown. Performance measures include three parts: (i) system performance index (average daily income、average waiting time per passenger), (ii) bus-stop performance index (average queue length、waiting time per passenger, average number of reneging passengers) and route performance index (number of passengers abroad, rate of success bus routes, average daily profit). Keywords: Stochastic System, Urban-Bus Simulation Software, Borland C++ Builder, Graphical User Interface.

並列關鍵字

Stochastic System ； Graphical User Interface ； Borland C++ Builder ； Urban-Bus Simulation Software

參考文獻

Rao, L., Owen, L., and Goldsman, D., “Development and Application of a Vaildation Framework for Traffic Simulation Models,” Proceedings of 1998 Winter Simulation Conference, 1079-1086.

劉方旗，市區公車排班與即時機動調度之研究---以新竹市公車為例，國立交通大學交通運輸研究所碩士論文，1998。