以時空資料探勘技術找出都會區交通路網瓶頸點的模式

Abstract

因為都市化及交通工具的普及，交通擁塞情形也越來越嚴重，尤其是都會區，許多研究因而提出來改善交通擁塞的問題，其中找到交通瓶頸點對於改善交通擁塞將會是非常有效且重要的議題。因為高速公路路網比都會區路網相對簡單的多且高速公路的瓶頸點大部分就位於閘道附近，所以大多數交通瓶頸點的相關研究都在高速公路。又因為都會區路網的瓶頸點是會隨著時間而改變的，所以找到都會區交通瓶頸點變成是一項非常困難但卻非常重要的任務。所以我們提出了一個時空交通瓶頸點探勘模組(Spatiotemporal Traffic Bottleneck Mining Model, STBM)利用資料探勘方式加上我們提出的三個瓶頸點特徵來找到都會區路網瓶頸點。我們的實驗設計在台北都會區，利用即時的計程車派遣系統(Taxi Dispatch System)來收集交通資訊，收集時間為2006/02到2007/03。從實驗結果可以看出，STBM的帄均準確率確實比傳統統計法的略高，幾乎有高達近八成。而且分布結果相當帄均也比較穩定。未來我們將會整合現有的STBM加上歷史的交通資訊以及即時交通資訊，發展一個新的交通瓶頸點及時預測系統，用以提供用路人或交通管理者更多及時有效的資訊。

The occurrence of traffic congestion has been increasing around world-wide as the result of the increasing of motorization, urbanization, population growth and changes in population density, especially in Urban Network; therefore, many researches are proposed to improve the traffic congestion; moreover, finding the traffic bottlenecks is the most important thing to improve the traffic congestion. As we know, freeway bottlenecks are always fixed and well known as gateway but the urban network bottlenecks may vary with spatial and temporal environment; therefore, finding out urban network bottleneck becomes a very difficult but very important mission. We propose a Spatiotemporal Traffic Bottleneck Mining Model (STBM) in this thesis to discover the urban network bottlenecks based on three heuristics we developed. In this thesis, STBM prototype model is implemented based on a real time LBS-based application to find out the Taipei urban network bottlenecks. Experimental results show that the average accuracy in workday of STBM is up to 80% and it‟s better than the traditional statistic model. In the near future, the STBM model could be implemented as a real time bottleneck detection and prediction system, which integrates the historical traffic patterns and real-time traffic information.