- 學位論文
混合車流交通安全評估熵參數之建立
Development of Entropy Safety Parameter for Mixed Traffic Flow
摘要
在台灣都會區中，摩托車或所謂機車之使用自幾十年前逐漸普及後即衍生出許多交通安全問題，原因在於機車騎士常自由地、機動地鑽行於車陣中，迫使原本單純的汽車車流轉化成為複雜的混合車流。本研究利用資訊熵理論發展出一可量化的交通安全評估參數，提供日後觀察機車對於混合車流安全影響之依據。此種交通安全評估熵參數於一計算時間視窗內之定義係以分類後的車輛速度為計算基礎，而車輛速度是利用相隔半秒之連續車流影像中之車輛中心點座標來計算。此評估參數主要是用以測量混合車流中各種車輛於車速上之波動性或一致性；當各車車速於同時段內之波動性越大時，車流即被認定是處於較不安全的狀態，波動性越小時則反之。然而，從研究中於台北市區內一距交叉路口上游44公尺之路段中所取得之資料發現，當熵參數與另一風險參數作比較驗證時，其與風險之正向線性關係只存在於號誌週期中之某特定區間。此風險參數之定義為偵測範圍內車輛碰撞機率與嚴重性之乘積；碰撞機率是以車速小於離碰撞時間（time to collision）門檻值之車量數佔總車輛數之百分比計算，而碰撞嚴重性則是此種車輛離碰撞時間與離碰撞時間門檻值間之平均差異。結果顯示本研究所定義之熵參數只對擁塞車流狀況有相當程度之安全評估能力，原因在於自由車流狀態下之高車速波動性未必會對車流造成危害。根據特定區間內所發現之熵與風險關係，本研究也相應地訂定了五種交通安全服務水準評估等級，做為未來混合車流中即時交通控制與管理之安全評估標準。
並列摘要
In Taiwanese metropolitan areas, the gradual popularization for the use of motorcycles or the so-called motorized scooters as one of the major transportation modes over the past decades has caused many safety concerns and forced traffic flow to become a mixed type when motorcyclists manoeuvre freely and haphazardly between vehicles on virtually any part of the urban roads. Before the effects of motorcycles in mixed traffic flow can be quantitatively observed, a safety parameter that involves the use of information entropy is first developed. The entropy parameter within a calculation time window is defined by computations of grouped vehicle speeds that are calculated by coordinates read on filmed traffic flow images at an interval of 0.5 seconds. It essentially measures the randomness or variation of vehicle speeds which is considered to indicate the relative safety level in mixed traffic flow. By relating and verifying entropy to risk, which is a traffic safety indicator that multiplies collision probability (percent of vehicles under a threshold time-to-collision value) by collision severity (average magnitude of time-to-collision value below the threshold), a positive linear trend has been identified during only specific periods of a signal cycle length for a road section 44 metres directly upstream of an intersection approach in Taipei City. Results indicate that entropy is descriptive of traffic safety levels only when congested flow conditions occur since great variations of vehicle speeds in free flow conditions do not necessarily cause collision hazards. By the discovered relation between entropy and risk, five scales of traffic safety level of service are established as standards for future real-time traffic control and management in mixed flow.